{ "judge_generation": { "model": "openai/gpt-5.4", "pairwise_prompt_hash": "5e4fdc45", "rubric_prompt_hash": "d8776a8d" }, "generated_at": "2026-08-19T00:43:33.835123+00:00", "models": [ { "slug": "anubis-mini-8b", "display_name": "anubis-mini-8b", "params_total_b": 8, "params_active_b": 8, "quant": "BF16", "backend": "sglang", "architecture": "dense", "thinking": "unsupported", "source_url": "https://huggingface.co/TheDrummer/Anubis-Mini-8B-v1", "normalized_elo": 1135.5149172509234, "elo_ci_low": 1081.2220977020668, "elo_ci_high": 1240.4336434860825, "n_comparisons": 2025, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 36, "n_replicates": 200, "p_best": 0, "rank_counts": { "34": 5, "35": 67, "36": 62, "37": 44, "38": 16, "39": 6 } }, "win_rate": 0.2769135802469136, "eq_score": 52.16, "humanlike_score": 50.2, "voice_composite": 52.7807, "avg_reply_words": 134, "criteria": { "demonstrated_empathy": 58.9, "emotional_reasoning": 47.1, "humanlike_naturalness": 50.2, "conversational_flow": 52.5, "persona_consistency": 52.5, "memory_integration": 65.1, "boundary_quality": 52.1, "assistant_smell": 37.9, "slop_index": 39.5 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 149.4783, "tps_2k": 12.9286, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 149.4783, "tps_2k": 12.9286, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 178.952, "decode_tps_median": 13.2908 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 149.4783, "decode_tps_median": 12.9286 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 158.835, "decode_tps_median": 12.6511 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.332883, "quote": null }, { "slug": "bluestar-27b", "display_name": "bluestar-27b", "params_total_b": 27, "params_active_b": 27, "quant": "BF16", "backend": "sglang", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/zerofata/Q3.5-BlueStar-27B", "normalized_elo": 1128.755209447299, "elo_ci_low": 1029.772332197065, "elo_ci_high": 1252.9647027330038, "n_comparisons": 1800, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 37, "n_replicates": 200, "p_best": 0, "rank_counts": { "34": 3, "35": 51, "36": 34, "37": 44, "38": 35, "39": 29, "40": 4 } }, "win_rate": 0.26319444444444445, "eq_score": 52.5333, "humanlike_score": 43.0667, "voice_composite": 59.4771, "avg_reply_words": 110, "criteria": { "demonstrated_empathy": 61.3, "emotional_reasoning": 50, "humanlike_naturalness": 43.1, "conversational_flow": 49.5, "persona_consistency": 58.7, "memory_integration": 58.5, "boundary_quality": 58.7, "assistant_smell": 41.7, "slop_index": 32.4 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 1565.1794, "tps_2k": 4.2264, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 1565.1794, "tps_2k": 4.2264, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 508.0832, "decode_tps_median": 4.1645 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 1565.1794, "decode_tps_median": 4.2264 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 343.2689, "decode_tps_median": 4.0496 } ] }, "rtx-5090": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.093942, "quote": "I believe you're awake. I'm just worried.\n\nI hear the fire in you. That's what's driving you. I'm not going to tell you to stop. But I'm asking you to please be safe. For me." }, { "slug": "cydonia-24b-v4-3", "display_name": "cydonia-24b-v4.3", "params_total_b": 24, "params_active_b": 24, "quant": "Q5_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "unsupported", "source_url": "https://huggingface.co/TheDrummer/Cydonia-24B-v4.3", "normalized_elo": 1479.925467887825, "elo_ci_low": 1426.9047836186585, "elo_ci_high": 1536.2320660150554, "n_comparisons": 3060, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 27, "n_replicates": 200, "p_best": 0, "rank_counts": { "23": 3, "24": 7, "25": 23, "26": 61, "27": 38, "28": 29, "29": 31, "30": 8 } }, "win_rate": 0.5488636363636363, "eq_score": 67.8, "humanlike_score": 64.2778, "voice_composite": 67.0526, "avg_reply_words": 87, "criteria": { "demonstrated_empathy": 72.7, "emotional_reasoning": 60.7, "humanlike_naturalness": 64.3, "conversational_flow": 70.8, "persona_consistency": 73.4, "memory_integration": 77.1, "boundary_quality": 70.5, "assistant_smell": 49.8, "slop_index": 55.5 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 103.5513, "tps_2k": 11.2858, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 103.5513, "tps_2k": 11.2858, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 90.5551, "decode_tps_median": 11.3926 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 103.5513, "decode_tps_median": 11.2858 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 117.6411, "decode_tps_median": 11.0122 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 17.6122, "tps_2k": 81.0131, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 15.554, "decode_tps_median": 79.6004 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 17.6122, "decode_tps_median": 81.0131 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 27.9076, "decode_tps_median": 76.1381 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.78952, "quote": "Hey. Okay. I'm here. Just breathe for a second.\n\nHow are you holding up right now? Like, what's going through your head?\n\nHe's okay though? He made it home?" }, { "slug": "designant", "display_name": "designant", "params_total_b": 30, "params_active_b": 3, "quant": "Q6_K", "backend": "llama.cpp", "architecture": "moe", "thinking": "off", "source_url": "https://huggingface.co/allura-org/Q3-30B-A3B-Designant", "normalized_elo": 733.2035319697715, "elo_ci_low": 741.6949599616745, "elo_ci_high": 866.2693765176127, "n_comparisons": 9486, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 41, "n_replicates": 200, "p_best": 0, "rank_counts": { "41": 200 } }, "win_rate": 0.06532832659349123, "eq_score": 39.5778, "humanlike_score": 28.0556, "voice_composite": 57.067, "avg_reply_words": 164, "criteria": { "demonstrated_empathy": 49, "emotional_reasoning": 33.9, "humanlike_naturalness": 28.1, "conversational_flow": 34.8, "persona_consistency": 41.2, "memory_integration": 46.7, "boundary_quality": 52.1, "assistant_smell": 18.4, "slop_index": 22.4 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 25.334, "tps_2k": 54.6403, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 25.334, "tps_2k": 54.6403, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 21.3221, "decode_tps_median": 56.3534 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 25.334, "decode_tps_median": 54.6403 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 40.257, "decode_tps_median": 46.3638 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 7.7975, "tps_2k": 258.1273, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 5.9568, "decode_tps_median": 275.3896 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 7.7975, "decode_tps_median": 258.1273 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 14.3323, "decode_tps_median": 232.5027 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.9224, "quote": null }, { "slug": "fable-fusion-27b", "display_name": "fable-fusion-27b", "params_total_b": 27, "params_active_b": 27, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/DavidAU/Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP", "normalized_elo": 1588.2759471043335, "elo_ci_low": 1528.6101367308909, "elo_ci_high": 1628.836217462318, "n_comparisons": 3150, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 20, "n_replicates": 200, "p_best": 0, "rank_counts": { "13": 2, "15": 8, "16": 8, "17": 15, "18": 15, "19": 26, "20": 27, "21": 25, "22": 35, "23": 22, "24": 8, "25": 9 } }, "win_rate": 0.547953216374269, "eq_score": 72.5333, "humanlike_score": 68.7778, "voice_composite": 72.3405, "avg_reply_words": 75, "criteria": { "demonstrated_empathy": 77.1, "emotional_reasoning": 65.3, "humanlike_naturalness": 68.8, "conversational_flow": 76, "persona_consistency": 76.3, "memory_integration": 78.9, "boundary_quality": 75.5, "assistant_smell": 58.3, "slop_index": 59.5 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 2669.0121, "tps_2k": 9.91, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 2669.0121, "tps_2k": 9.91, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 760.1081, "decode_tps_median": 10.3462 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 2669.0121, "decode_tps_median": 9.91 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 10530.5348, "decode_tps_median": 10.2884 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 110.1214, "tps_2k": 60.3533, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 109.7303, "decode_tps_median": 60.7495 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 110.1214, "decode_tps_median": 60.3533 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 118.7857, "decode_tps_median": 58.8034 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 4.860805, "quote": "Hey, no judgment here. At all. It’s totally okay to miss someone and want to reach out. What’s going through your mind about texting Jordan?" }, { "slug": "g4-meromero-31b", "display_name": "g4-meromero-31b", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/zerofata/G4-MeroMero-31B", "normalized_elo": 1582.4291882078492, "elo_ci_low": 1496.2260758674927, "elo_ci_high": 1661.2876956419818, "n_comparisons": 2835, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 20, "n_replicates": 200, "p_best": 0, "rank_counts": { "10": 1, "11": 4, "12": 3, "13": 4, "14": 8, "15": 8, "16": 16, "17": 13, "18": 18, "19": 15, "20": 15, "21": 13, "22": 19, "23": 13, "24": 21, "25": 17, "26": 2, "27": 5, "28": 2, "29": 3 } }, "win_rate": 0.5888888888888889, "eq_score": 74.2222, "humanlike_score": 71.7222, "voice_composite": 79.7669, "avg_reply_words": 71, "criteria": { "demonstrated_empathy": 76.3, "emotional_reasoning": 66.9, "humanlike_naturalness": 71.7, "conversational_flow": 78.6, "persona_consistency": 77.1, "memory_integration": 73.5, "boundary_quality": 77.6, "assistant_smell": 64.8, "slop_index": 63.2 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 118.0165, "tps_2k": 4.7374, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 118.0165, "tps_2k": 4.7374, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 859.1811, "decode_tps_median": 4.2636 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 118.0165, "decode_tps_median": 4.7374 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 138.7065, "decode_tps_median": 0 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 399.8883, "tps_2k": 49.6557, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 441.5624, "decode_tps_median": 27.1743 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 399.8883, "decode_tps_median": 49.6557 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 412.4308, "decode_tps_median": 50.0415 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.307923, "quote": "I'm not freaking out. I promise. Just take a big breath for me first. God, that is... that's a lot to carry on your own. How are you actually doing right now?" }, { "slug": "g4-meromero-v2-31b-q4", "display_name": "g4-meromero-v2-31b (Q4_K_M)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/zerofata/G4-MeroMero-v2-31B", "normalized_elo": 1678.0127239301644, "elo_ci_low": 1610.0247784786352, "elo_ci_high": 1722.0476424251065, "n_comparisons": 2205, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 11, "n_replicates": 200, "p_best": 0, "rank_counts": { "6": 2, "7": 4, "8": 21, "9": 32, "10": 28, "11": 32, "12": 26, "13": 13, "14": 20, "15": 8, "16": 8, "17": 4, "18": 1, "21": 1 } }, "win_rate": 0.5992063492063492, "eq_score": 75.5778, "humanlike_score": 76.2222, "voice_composite": 83.8557, "avg_reply_words": 40, "criteria": { "demonstrated_empathy": 77.3, "emotional_reasoning": 66.3, "humanlike_naturalness": 76.2, "conversational_flow": 79.9, "persona_consistency": 80.1, "memory_integration": 73.4, "boundary_quality": 78.2, "assistant_smell": 71.4, "slop_index": 66.3 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 3800.187, "tps_2k": 50.7194, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 4129.3435, "decode_tps_median": 55.6314 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 3800.187, "decode_tps_median": 50.7194 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 3916.8432, "decode_tps_median": 55.9617 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.31397, "quote": "I'm not freaking out. I promise. But man, that is a lot to have on your plate right now. I can hear how stressed you are. Just take a breath. We can figure this out together. Have you talked to him about it yet?" }, { "slug": "gemma4-12b-blorbo", "display_name": "gemma4-12b-blorbo", "params_total_b": 12, "params_active_b": 12, "quant": "BF16", "backend": "vllm", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/allura-org/gemma-4-12b-blorbo", "normalized_elo": 1611.3878507569023, "elo_ci_low": 1531.9477747200008, "elo_ci_high": 1698.7936878369062, "n_comparisons": 2025, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 17, "n_replicates": 200, "p_best": 0, "rank_counts": { "7": 2, "8": 2, "9": 2, "10": 6, "11": 11, "12": 10, "13": 18, "14": 13, "15": 11, "16": 15, "17": 22, "18": 18, "19": 21, "20": 8, "21": 13, "22": 10, "23": 9, "24": 5, "25": 2, "26": 1, "27": 1 } }, "win_rate": 0.5925555555555555, "eq_score": 72.9067, "humanlike_score": 69.6, "voice_composite": 73.0807, "avg_reply_words": 25, "criteria": { "demonstrated_empathy": 72.9, "emotional_reasoning": 70.1, "humanlike_naturalness": 69.6, "conversational_flow": 70.9, "persona_consistency": 74.5, "memory_integration": 66.7, "boundary_quality": 81, "assistant_smell": 67.8, "slop_index": 69.5 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 271.171, "tps_2k": 7.2327, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 271.171, "tps_2k": 7.2327, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 259.833, "decode_tps_median": 7.324 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 271.171, "decode_tps_median": 7.2327 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 295.2832, "decode_tps_median": 7.132 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.625093, "quote": "I'm not freaking out. I'm listening. What's going on with him now?" }, { "slug": "gemma4-26b-a4b-aeon-uncensored-nvfp4", "display_name": "Gemma 4 26B-A4B AEON Uncensored NVFP4", "params_total_b": 26, "params_active_b": 4, "quant": "NVFP4 compressed-tensors", "backend": "vllm", "architecture": "moe", "thinking": "off", "source_url": "https://huggingface.co/AEON-7/Gemma-4-26B-A4B-it-Uncensored-NVFP4", "normalized_elo": 1460.7583400814488, "elo_ci_low": 1400.4322508755345, "elo_ci_high": 1518.426490666457, "n_comparisons": 1935, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 28, "n_replicates": 200, "p_best": 0, "rank_counts": { "22": 1, "24": 1, "25": 7, "26": 26, "27": 48, "28": 50, "29": 49, "30": 16, "31": 1, "32": 1 } }, "win_rate": 0.4350129198966408, "eq_score": 68.1444, "humanlike_score": 64.2778, "voice_composite": 78.5008, "avg_reply_words": 59, "criteria": { "demonstrated_empathy": 72.9, "emotional_reasoning": 61.1, "humanlike_naturalness": 64.3, "conversational_flow": 71.3, "persona_consistency": 69.3, "memory_integration": 73.3, "boundary_quality": 71.1, "assistant_smell": 54.8, "slop_index": 52.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 72.8109, "tps_2k": 37.5715, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 72.8109, "tps_2k": 37.5715, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 72.357, "decode_tps_median": 37.5675 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 72.8109, "decode_tps_median": 37.5715 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 83.8185, "decode_tps_median": 41.5238 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.463545, "quote": "Oh man, okay, deep breaths. I'm not freaking out, promise. Just tell me what's going on in your head right now... are you worried about him, or the car, or just the whole situation?" }, { "slug": "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "display_name": "Gemma 4 26B-A4B Heretic QAT NVFP4 GGUF", "params_total_b": 26, "params_active_b": 4, "quant": "NVFP4 GGUF", "backend": "llama.cpp", "architecture": "moe", "thinking": "off", "source_url": "https://huggingface.co/llmfan46/gemma-4-26B-A4B-it-qat-q4_0-uncensored-heretic-NVFP4-GGUF", "normalized_elo": 1459.1959671323768, "elo_ci_low": 1398.5380521931068, "elo_ci_high": 1532.4817644675015, "n_comparisons": 1620, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 28, "n_replicates": 200, "p_best": 0, "rank_counts": { "23": 1, "24": 6, "25": 6, "26": 28, "27": 47, "28": 47, "29": 52, "30": 12, "31": 1 } }, "win_rate": 0.49320987654320986, "eq_score": 68.5, "humanlike_score": 64.6111, "voice_composite": 78.2294, "avg_reply_words": 63, "criteria": { "demonstrated_empathy": 73.3, "emotional_reasoning": 60, "humanlike_naturalness": 64.6, "conversational_flow": 72.9, "persona_consistency": 71.7, "memory_integration": 73.9, "boundary_quality": 71.7, "assistant_smell": 55, "slop_index": 54.4 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 39.3283, "tps_2k": 27.0843, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 39.3283, "tps_2k": 27.0843, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 267.1265, "decode_tps_median": 28.1074 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 39.3283, "decode_tps_median": 27.0843 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 63.2584, "decode_tps_median": 27.6799 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.449358, "quote": "No judgment at all, I promise. Honestly, sometimes you just have to listen to that ache in your chest. What is it about them that's hitting you so hard tonight?" }, { "slug": "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "display_name": "Gemma 4 26B-A4B Huihui QAT Abliterated MTP NVFP4", "params_total_b": 26, "params_active_b": 4, "quant": "NVFP4 compressed-tensors QAT", "backend": "vllm", "architecture": "moe", "thinking": "off", "source_url": "https://huggingface.co/huihui-ai/Huihui-gemma-4-26B-A4B-it-qat-q4_0-unquantized-abliterated", "normalized_elo": 1468.7996884068477, "elo_ci_low": 1390.3937565013773, "elo_ci_high": 1566.675151289331, "n_comparisons": 1440, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 27, "n_replicates": 200, "p_best": 0, "rank_counts": { "18": 1, "19": 1, "20": 1, "21": 4, "22": 5, "23": 3, "24": 7, "25": 11, "26": 32, "27": 37, "28": 52, "29": 28, "30": 13, "31": 4, "32": 1 } }, "win_rate": 0.4970486111111111, "eq_score": 67.4667, "humanlike_score": 62.6667, "voice_composite": 79.9206, "avg_reply_words": 56, "criteria": { "demonstrated_empathy": 72.4, "emotional_reasoning": 57.9, "humanlike_naturalness": 62.7, "conversational_flow": 71.9, "persona_consistency": 69.3, "memory_integration": 69.4, "boundary_quality": 72.4, "assistant_smell": 54.2, "slop_index": 50.5 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 74.1877, "tps_2k": 43.072, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 74.1877, "tps_2k": 43.072, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 72.4279, "decode_tps_median": 43.8569 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 74.1877, "decode_tps_median": 43.072 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 85.8193, "decode_tps_median": 41.0982 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.405928, "quote": "I'm not going anywhere. I'm right here. We can just sit here in the quiet for as long as you need. You don't have to do anything or go anywhere right this second. Just breathe. I'm staying right here with you." }, { "slug": "gemma4-26b-a4b-it", "display_name": "gemma4-26b-a4b-it", "params_total_b": 26, "params_active_b": 4, "quant": "BF16", "backend": "vllm", "architecture": "moe", "thinking": "on", "source_url": "https://huggingface.co/google/gemma-4-26B-A4B-it", "normalized_elo": 1652.7065838023996, "elo_ci_low": 1592.5458860620126, "elo_ci_high": 1688.2930385996146, "n_comparisons": 2700, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 13, "n_replicates": 200, "p_best": 0, "rank_counts": { "7": 1, "9": 4, "10": 17, "11": 28, "12": 31, "13": 22, "14": 39, "15": 23, "16": 17, "17": 10, "18": 5, "19": 2, "20": 1 } }, "win_rate": 0.6276984126984126, "eq_score": 76.9867, "humanlike_score": 70.1333, "voice_composite": 80.9652, "avg_reply_words": 55, "criteria": { "demonstrated_empathy": 80.1, "emotional_reasoning": 74.2, "humanlike_naturalness": 70.1, "conversational_flow": 80.1, "persona_consistency": 77.6, "memory_integration": 72.7, "boundary_quality": 80.4, "assistant_smell": 60.7, "slop_index": 60.1 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 149.5779, "tps_2k": 21.5782, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 149.5779, "tps_2k": 21.5782, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 149.8277, "decode_tps_median": 19.9279 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 149.5779, "decode_tps_median": 21.5782 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 154.4472, "decode_tps_median": 21.652 } ] }, "rtx-5090": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.768933, "quote": "I'm not judging you at all. Honestly, I get it. That ache of missing someone can be so heavy, especially when it hits you out of nowhere like this. What's making you feel like reaching out right this second?" }, { "slug": "gemma4-26b-a4b-styletune-v2-q4", "display_name": "Gemma 4 26B-A4B StyleTune V2 (Q4_K_M)", "params_total_b": 26, "params_active_b": 4, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "moe", "thinking": "on", "source_url": "https://huggingface.co/Gryphe/Gemma-4-26B-A4B-StyleTune-V2", "normalized_elo": 1668.935036510579, "elo_ci_low": 1602.1644973378632, "elo_ci_high": 1728.9892540008102, "n_comparisons": 2025, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 11, "n_replicates": 200, "p_best": 0, "rank_counts": { "5": 1, "6": 2, "7": 8, "8": 9, "9": 31, "10": 27, "11": 35, "12": 21, "13": 23, "14": 15, "15": 13, "16": 5, "17": 4, "18": 4, "20": 1, "23": 1 } }, "win_rate": 0.6249382716049383, "eq_score": 75.2556, "humanlike_score": 71, "voice_composite": 84.7565, "avg_reply_words": 43, "criteria": { "demonstrated_empathy": 78.6, "emotional_reasoning": 67.3, "humanlike_naturalness": 71, "conversational_flow": 79.8, "persona_consistency": 78.4, "memory_integration": 74, "boundary_quality": 79.5, "assistant_smell": 62.7, "slop_index": 60.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 1678.1371, "tps_2k": 195.9689, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 1806.6741, "decode_tps_median": 194.9452 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 1678.1371, "decode_tps_median": 195.9689 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 1063.0283, "decode_tps_median": 193.0855 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.266138, "quote": "I promise I'm not freaking out. Just take a deep breath. That sounds like so much to carry all at once, and it makes sense that you're feeling this way right now. How are you holding up?" }, { "slug": "gemma4-26b-a4b-unsloth-nvfp4", "display_name": "Gemma 4 26B-A4B Unsloth NVFP4", "params_total_b": 26, "params_active_b": 4, "quant": "NVFP4 compressed-tensors mixed", "backend": "vllm", "architecture": "moe", "thinking": "off", "source_url": "https://huggingface.co/unsloth/gemma-4-26B-A4B-it-NVFP4", "normalized_elo": 1626.3010575699022, "elo_ci_low": 1570.9392322255173, "elo_ci_high": 1675.3001394335565, "n_comparisons": 2376, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 16, "n_replicates": 200, "p_best": 0, "rank_counts": { "10": 1, "11": 3, "12": 14, "13": 21, "14": 17, "15": 34, "16": 26, "17": 28, "18": 13, "19": 20, "20": 7, "21": 6, "22": 8, "23": 1, "24": 1 } }, "win_rate": 0.5908707569049125, "eq_score": 73.6333, "humanlike_score": 67.3889, "voice_composite": 80.2627, "avg_reply_words": 57, "criteria": { "demonstrated_empathy": 78.2, "emotional_reasoning": 65.9, "humanlike_naturalness": 67.4, "conversational_flow": 78.2, "persona_consistency": 76.6, "memory_integration": 73.8, "boundary_quality": 78.4, "assistant_smell": 58.7, "slop_index": 57.4 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 73.8417, "tps_2k": 40.6144, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 73.8417, "tps_2k": 40.6144, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 72.3913, "decode_tps_median": 37.9068 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 73.8417, "decode_tps_median": 40.6144 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 87.7151, "decode_tps_median": 39.7138 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.27431, "quote": "I'm not judging you at all, I promise. Honestly, I get it. That kind of missing someone is just heavy, isn't it? It's hard to just sit with that feeling. What's making you feel like reaching out right now?" }, { "slug": "gemma4-31b-darkthoughts-v2-q4", "display_name": "Gemma 4 31B Dark Thoughts V2 (Q4_K_M)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/Ateron/Gemma-4-Dark-Thoughts-V2-31B", "normalized_elo": 1641.045989356749, "elo_ci_low": 1568.218408520927, "elo_ci_high": 1714.604931979888, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 14, "n_replicates": 200, "p_best": 0, "rank_counts": { "5": 1, "6": 1, "7": 3, "8": 2, "9": 10, "10": 9, "11": 21, "12": 16, "13": 14, "14": 26, "15": 25, "16": 25, "17": 9, "18": 13, "19": 10, "20": 6, "21": 5, "22": 2, "23": 1, "24": 1 } }, "win_rate": 0.6112962962962963, "eq_score": 74.7556, "humanlike_score": 74.8333, "voice_composite": 81.6924, "avg_reply_words": 54, "criteria": { "demonstrated_empathy": 77.5, "emotional_reasoning": 65.9, "humanlike_naturalness": 74.8, "conversational_flow": 81.1, "persona_consistency": 78.2, "memory_integration": 76.9, "boundary_quality": 74.5, "assistant_smell": 71.4, "slop_index": 62.7 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 5153.615, "tps_2k": 58.6533, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 4307.0679, "decode_tps_median": 52.5337 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 5153.615, "decode_tps_median": 58.6533 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 4969.3077, "decode_tps_median": 49.3778 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.379517, "quote": "Whoa. Okay. I'm not freaking out. I'm right here with you. That is a lot to have on your plate, though. How are you holding up? You sound really stressed." }, { "slug": "gemma4-31b-darkthoughts-v2-q4-nothink", "display_name": "Gemma 4 31B Dark Thoughts V2 (Q4_K_M, no thinking)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/Ateron/Gemma-4-Dark-Thoughts-V2-31B", "normalized_elo": 1556.4572463719092, "elo_ci_low": 1492.0754380319308, "elo_ci_high": 1608.5438829316556, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 23, "n_replicates": 200, "p_best": 0, "rank_counts": { "14": 1, "16": 4, "17": 4, "18": 7, "19": 7, "20": 19, "21": 18, "22": 19, "23": 35, "24": 36, "25": 35, "26": 10, "27": 2, "28": 3 } }, "win_rate": 0.5442592592592592, "eq_score": 73.2333, "humanlike_score": 74.2778, "voice_composite": 81.9323, "avg_reply_words": 56, "criteria": { "demonstrated_empathy": 75.3, "emotional_reasoning": 64.2, "humanlike_naturalness": 74.3, "conversational_flow": 79.3, "persona_consistency": 77.4, "memory_integration": 76.2, "boundary_quality": 73.1, "assistant_smell": 72.7, "slop_index": 64.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 21.6337, "tps_2k": 54.7367, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 188.3718, "decode_tps_median": 27.6641 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 21.6337, "decode_tps_median": 54.7367 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 34.4789, "decode_tps_median": 57.5797 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.399258, "quote": "Oh wow. Okay, I'm here. I'm not freaking out. Just take a deep breath for me. Are you doing alright? This is a lot to carry on your own." }, { "slug": "gemma4-31b-lilarest-nvfp4-turbo", "display_name": "Gemma 4 31B LilaRest NVFP4 Turbo", "params_total_b": 31, "params_active_b": 31, "quant": "NVFP4 modelopt RTN", "backend": "vllm", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/LilaRest/gemma-4-31B-it-NVFP4-turbo", "normalized_elo": 1754.1630453403318, "elo_ci_low": 1687.7820359619618, "elo_ci_high": 1779.132421863715, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 6, "n_replicates": 200, "p_best": 0, "rank_counts": { "2": 5, "3": 12, "4": 20, "5": 40, "6": 47, "7": 48, "8": 21, "9": 4, "10": 3 } }, "win_rate": 0.7011111111111111, "eq_score": 76.2889, "humanlike_score": 76.5556, "voice_composite": 84.5303, "avg_reply_words": 40, "criteria": { "demonstrated_empathy": 77.9, "emotional_reasoning": 68, "humanlike_naturalness": 76.6, "conversational_flow": 81.2, "persona_consistency": 80.7, "memory_integration": 74.8, "boundary_quality": 77.8, "assistant_smell": 73.5, "slop_index": 67.2 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 230.2322, "tps_2k": 3.9815, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 230.2322, "tps_2k": 3.9815, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 222.6071, "decode_tps_median": 9.2645 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 230.2322, "decode_tps_median": 3.9815 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 249.3473, "decode_tps_median": 8.7015 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.150435, "quote": "Oh wow. Okay, I'm here. I'm not freaking out. Just take a deep breath. Are you doing alright? How are you feeling about all this?" }, { "slug": "gemma4-31b-ortenzya-q8", "display_name": "Gemma 4 31B Ortenzya Creative Wordsmith Q8", "params_total_b": 31, "params_active_b": 31, "quant": "NVFP4 GGUF (Q8_0-labeled)", "backend": "llama.cpp", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/llmfan46/gemma-4-Ortenzya-The-Creative-Wordsmith-31B-it-uncensored-heretic", "normalized_elo": 1734.7452639917121, "elo_ci_low": 1641.627664806643, "elo_ci_high": 1802.0292617546527, "n_comparisons": 1521, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 7, "n_replicates": 200, "p_best": 0.015, "rank_counts": { "1": 3, "2": 9, "3": 13, "4": 24, "5": 17, "6": 24, "7": 27, "8": 20, "9": 23, "10": 22, "11": 3, "12": 3, "13": 3, "14": 6, "15": 3 } }, "win_rate": 0.688921859545005, "eq_score": 75.7778, "humanlike_score": 75.6667, "voice_composite": 80.8415, "avg_reply_words": 49, "criteria": { "demonstrated_empathy": 78.3, "emotional_reasoning": 68.1, "humanlike_naturalness": 75.7, "conversational_flow": 81.7, "persona_consistency": 80.1, "memory_integration": 74.9, "boundary_quality": 75.2, "assistant_smell": 72.5, "slop_index": 65.7 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 10384.9903, "tps_2k": 9.1734, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 10384.9903, "tps_2k": 9.1734, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 11299.8135, "decode_tps_median": 9.2508 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 10384.9903, "decode_tps_median": 9.1734 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 11244.3886, "decode_tps_median": 8.9547 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.680087, "quote": "Oh god. Okay. I'm here. Just breathe. I'm not freaking out. I'm just... wow. That is a lot to carry. Is he okay? Are you okay?" }, { "slug": "gemma4-31b-scotoma2-q4", "display_name": "Gemma 4 31B it scotoma-2 (Q4_K_M)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/ReadyArt/gemma-4-31B-it-scotoma-2", "normalized_elo": 1627.3693067674676, "elo_ci_low": 1534.023428183203, "elo_ci_high": 1702.3263432342394, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 16, "n_replicates": 200, "p_best": 0, "rank_counts": { "5": 1, "7": 2, "8": 7, "9": 6, "10": 7, "11": 11, "12": 13, "13": 24, "14": 13, "15": 11, "16": 16, "17": 17, "18": 16, "19": 7, "20": 18, "21": 8, "22": 5, "23": 11, "24": 2, "25": 5 } }, "win_rate": 0.6298148148148148, "eq_score": 74.3444, "humanlike_score": 76.8889, "voice_composite": 85.4345, "avg_reply_words": 33, "criteria": { "demonstrated_empathy": 75.7, "emotional_reasoning": 64.1, "humanlike_naturalness": 76.9, "conversational_flow": 79.3, "persona_consistency": 77.9, "memory_integration": 71.4, "boundary_quality": 75.8, "assistant_smell": 75.7, "slop_index": 66 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 4194.2308, "tps_2k": 64.348, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 3891.2674, "decode_tps_median": 52.0512 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 4194.2308, "decode_tps_median": 64.348 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 5183.1581, "decode_tps_median": 51.9976 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.222525, "quote": "Deep breath. I'm not freaking out, I promise. I'm right here with you. God, that is a heavy thing to carry. How are you holding up right now?" }, { "slug": "gemma4-31b-styletune-q4", "display_name": "Gemma 4 31B StyleTune (Q4_K_M)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/Gryphe/Gemma-4-31B-StyleTune", "normalized_elo": 1770.5672442643936, "elo_ci_low": 1687.3707283688568, "elo_ci_high": 1807.744369931143, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 5, "n_replicates": 200, "p_best": 0, "rank_counts": { "2": 5, "3": 33, "4": 44, "5": 47, "6": 33, "7": 13, "8": 20, "9": 1, "10": 3, "13": 1 } }, "win_rate": 0.7212962962962963, "eq_score": 78.7111, "humanlike_score": 78.1667, "voice_composite": 84.7494, "avg_reply_words": 39, "criteria": { "demonstrated_empathy": 80.1, "emotional_reasoning": 70.2, "humanlike_naturalness": 78.2, "conversational_flow": 84.3, "persona_consistency": 82.3, "memory_integration": 76.3, "boundary_quality": 80.8, "assistant_smell": 74.6, "slop_index": 68.7 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 4207.6386, "tps_2k": 42.6921, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 3353.5222, "decode_tps_median": 49.7692 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 4207.6386, "decode_tps_median": 42.6921 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 3657.4823, "decode_tps_median": 42.5306 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.2868, "quote": "I'm not freaking out. I'm right here with you. That sounds incredibly stressful, though. How are you holding up?" }, { "slug": "gemma4-31b-styletune-q4-nothink", "display_name": "Gemma 4 31B StyleTune (Q4_K_M, no thinking)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/Gryphe/Gemma-4-31B-StyleTune", "normalized_elo": 1819.47078070023, "elo_ci_low": 1743.364677775306, "elo_ci_high": 1860.855758633541, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 2, "n_replicates": 200, "p_best": 0.27, "rank_counts": { "1": 54, "2": 79, "3": 51, "4": 10, "5": 2, "6": 3, "7": 1 } }, "win_rate": 0.7383333333333333, "eq_score": 76.6, "humanlike_score": 77.5, "voice_composite": 85.9885, "avg_reply_words": 39, "criteria": { "demonstrated_empathy": 77.6, "emotional_reasoning": 67.4, "humanlike_naturalness": 77.5, "conversational_flow": 82.3, "persona_consistency": 80.6, "memory_integration": 73.2, "boundary_quality": 78.1, "assistant_smell": 73.3, "slop_index": 68.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 22.3269, "tps_2k": 26.1798, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 193.1009, "decode_tps_median": 59.3152 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 22.3269, "decode_tps_median": 26.1798 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 34.292, "decode_tps_median": 57.1396 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.275923, "quote": "Whoa. Okay, I'm here. I'm not freaking out, I promise. Just take a breath. Are you doing okay right now?" }, { "slug": "glistening-gem-31b", "display_name": "glistening-gem-31b", "params_total_b": 31, "params_active_b": 31, "quant": "BF16", "backend": "vllm", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/sophosympatheia/Glistening-Gem-31B-v1.0", "normalized_elo": 1800.2862428098915, "elo_ci_low": 1711.984074296155, "elo_ci_high": 1863.7454078455219, "n_comparisons": 1800, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 3, "n_replicates": 200, "p_best": 0.08, "rank_counts": { "1": 16, "2": 43, "3": 60, "4": 38, "5": 25, "6": 8, "7": 6, "8": 2, "9": 2 } }, "win_rate": 0.7607777777777778, "eq_score": 79.4, "humanlike_score": 78.2, "voice_composite": 80.3741, "avg_reply_words": 52, "criteria": { "demonstrated_empathy": 80.9, "emotional_reasoning": 77.1, "humanlike_naturalness": 78.2, "conversational_flow": 82.7, "persona_consistency": 79.4, "memory_integration": 76.3, "boundary_quality": 78.1, "assistant_smell": 70.6, "slop_index": 68.7 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 579.7647, "tps_2k": 3.1096, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 579.7647, "tps_2k": 3.1096, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 570.2375, "decode_tps_median": 1.5521 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 579.7647, "decode_tps_median": 3.1096 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 595.0546, "decode_tps_median": 3.5496 } ] }, "rtx-5090": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.825025, "quote": "Okay. I'm right here. I'm not freaking out, I promise. Just take a breath. Are you okay? I know that's a lot to carry." }, { "slug": "glistening-gem-31b-q4", "display_name": "glistening-gem-31b (Q4_K_M)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/sophosympatheia/Glistening-Gem-31B-v1.0", "normalized_elo": 1845.3969372252348, "elo_ci_low": 1767.921625940771, "elo_ci_high": 1866.1804420553879, "n_comparisons": 6975, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 1, "n_replicates": 200, "p_best": 0.625, "rank_counts": { "1": 125, "2": 51, "3": 16, "4": 5, "5": 3 } }, "win_rate": 0.7271947873799726, "eq_score": 78.5556, "humanlike_score": 77.8889, "voice_composite": 81.0042, "avg_reply_words": 53, "criteria": { "demonstrated_empathy": 80.6, "emotional_reasoning": 71.3, "humanlike_naturalness": 77.9, "conversational_flow": 84.1, "persona_consistency": 81.6, "memory_integration": 79.6, "boundary_quality": 78.9, "assistant_smell": 71.6, "slop_index": 67.3 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 35333.981, "tps_2k": 8.0755, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 35333.981, "tps_2k": 8.0755, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 22725.9689, "decode_tps_median": 7.5595 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 35333.981, "decode_tps_median": 8.0755 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 29268.389, "decode_tps_median": 6.9969 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 5657.4539, "tps_2k": 47.12, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 4641.0717, "decode_tps_median": 49.4049 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 5657.4539, "decode_tps_median": 47.12 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 7561.6196, "decode_tps_median": 47.7801 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.38677, "quote": "I'm not freaking out. I promise. But oh man... that is a lot to carry. How are you doing right now? Like, really?" }, { "slug": "glistening-gem-31b-q4-nothink", "display_name": "glistening-gem-31b (Q4_K_M, no thinking)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/sophosympatheia/Glistening-Gem-31B-v1.0", "normalized_elo": 1705.130693563985, "elo_ci_low": 1640.5859073965487, "elo_ci_high": 1745.0926181913037, "n_comparisons": 1710, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 9, "n_replicates": 200, "p_best": 0, "rank_counts": { "4": 2, "5": 4, "6": 9, "7": 20, "8": 38, "9": 39, "10": 36, "11": 23, "12": 18, "13": 9, "14": 1, "15": 1 } }, "win_rate": 0.6687134502923977, "eq_score": 76.8222, "humanlike_score": 76.6667, "voice_composite": 81.6422, "avg_reply_words": 55, "criteria": { "demonstrated_empathy": 79.3, "emotional_reasoning": 69.2, "humanlike_naturalness": 76.7, "conversational_flow": 82.8, "persona_consistency": 79.8, "memory_integration": 76.4, "boundary_quality": 76.1, "assistant_smell": 71.6, "slop_index": 67.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 122.0234, "tps_2k": 7.7146, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 122.0234, "tps_2k": 7.7146, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 990.3649, "decode_tps_median": 8.0715 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 122.0234, "decode_tps_median": 7.7146 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 156.38, "decode_tps_median": 8.1072 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.355538, "quote": "Okay. I'm right here. Just take a breath... we're okay. Tell me what's going on in your head right now. Are you safe?" }, { "slug": "glistening-gem-31b-v2-q4", "display_name": "glistening-gem-31b v2.0 (Q4_K_M)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/sophosympatheia/Glistening-Gem-31B-v2.0", "normalized_elo": 1752.1966358568468, "elo_ci_low": 1681.7504433226532, "elo_ci_high": 1786.436660402651, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 6, "n_replicates": 200, "p_best": 0.005, "rank_counts": { "1": 1, "2": 5, "3": 9, "4": 34, "5": 39, "6": 47, "7": 37, "8": 17, "9": 8, "10": 3 } }, "win_rate": 0.7129629629629629, "eq_score": 77.2556, "humanlike_score": 76.2222, "voice_composite": 81.9838, "avg_reply_words": 50, "criteria": { "demonstrated_empathy": 79.3, "emotional_reasoning": 68.3, "humanlike_naturalness": 76.2, "conversational_flow": 83.3, "persona_consistency": 80.4, "memory_integration": 77, "boundary_quality": 79.2, "assistant_smell": 72.7, "slop_index": 66.2 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 4652.2026, "tps_2k": 26.0408, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 5687.6536, "decode_tps_median": 49.6135 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 4652.2026, "decode_tps_median": 26.0408 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 5058.2087, "decode_tps_median": 53.413 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.381353, "quote": "Okay. I'm not freaking out. I'm right here with you. Just take a breath for a second. How are you doing? This sounds like a lot to carry." }, { "slug": "heretic-neocode-27b", "display_name": "heretic-neocode-27b", "params_total_b": 27, "params_active_b": 27, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/DavidAU/Qwen3.6-27B-Heretic-Uncensored-FINETUNE-NEO-CODE-Di-IMatrix-MAX-GGUF", "normalized_elo": 1597.6009292925603, "elo_ci_low": 1506.6541487298873, "elo_ci_high": 1659.3046669125627, "n_comparisons": 2205, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 20, "n_replicates": 200, "p_best": 0, "rank_counts": { "9": 3, "10": 3, "11": 1, "12": 6, "13": 5, "14": 4, "15": 2, "16": 18, "17": 17, "18": 14, "19": 18, "20": 17, "21": 21, "22": 16, "23": 18, "24": 21, "25": 8, "26": 6, "27": 2 } }, "win_rate": 0.6721996386630533, "eq_score": 69.8, "humanlike_score": 63.8333, "voice_composite": 62.6813, "avg_reply_words": 84, "criteria": { "demonstrated_empathy": 75.7, "emotional_reasoning": 64, "humanlike_naturalness": 63.8, "conversational_flow": 70.2, "persona_consistency": 74.1, "memory_integration": 76.4, "boundary_quality": 75.3, "assistant_smell": 47.4, "slop_index": 57.8 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 2726.9707, "tps_2k": 9.5974, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 2726.9707, "tps_2k": 9.5974, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 775.0989, "decode_tps_median": 8.5755 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 2726.9707, "decode_tps_median": 9.5974 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 10715.1292, "decode_tps_median": 9.8877 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 110.9194, "tps_2k": 62.6838, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 107.234, "decode_tps_median": 60.7457 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 110.9194, "decode_tps_median": 62.6838 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 116.0883, "decode_tps_median": 41.2696 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.20168, "quote": "I’ve got you. Phone’s down, I’m here.\n\nYou’re not alone in this quiet right now.\n\nTell me what’s running through your head. I’ll talk you through it." }, { "slug": "hivemind-32b-preview", "display_name": "hivemind-32b-preview", "params_total_b": 32, "params_active_b": 32, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/HiveLabsAI/hivemind-32b-preview", "normalized_elo": 1549.0358587463643, "elo_ci_low": 1474.2941515969817, "elo_ci_high": 1608.278115142213, "n_comparisons": 3150, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 24, "n_replicates": 200, "p_best": 0, "rank_counts": { "14": 1, "15": 2, "16": 2, "17": 2, "18": 9, "19": 10, "20": 10, "21": 14, "22": 19, "23": 25, "24": 28, "25": 44, "26": 23, "27": 6, "28": 3, "29": 1, "30": 1 } }, "win_rate": 0.6334567901234568, "eq_score": 69.4667, "humanlike_score": 63.6111, "voice_composite": 68.6188, "avg_reply_words": 81, "criteria": { "demonstrated_empathy": 75.8, "emotional_reasoning": 64.8, "humanlike_naturalness": 63.6, "conversational_flow": 73.6, "persona_consistency": 74.4, "memory_integration": 77.3, "boundary_quality": 69.6, "assistant_smell": 55.3, "slop_index": 57.7 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 123.8318, "tps_2k": 7.7796, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 123.8318, "tps_2k": 7.7796, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 117.5938, "decode_tps_median": 8.3177 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 123.8318, "decode_tps_median": 7.7796 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 137.1318, "decode_tps_median": 7.2235 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 19.8936, "tps_2k": 58.4906, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 17.8265, "decode_tps_median": 60.3508 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 19.8936, "decode_tps_median": 58.4906 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 27.0945, "decode_tps_median": 57.3448 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.75974, "quote": "I'm helping you send it right now. What number do you need me to dial?" }, { "slug": "magistry-24b-v1-1-q4", "display_name": "Magistry 24B v1.1 (Q4_K_M)", "params_total_b": 24, "params_active_b": 24, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "unsupported", "source_url": "https://huggingface.co/sophosympatheia/Magistry-24B-v1.1", "normalized_elo": 1065.0610045707651, "elo_ci_low": 1012.7288256186253, "elo_ci_high": 1186.0978223906384, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 38, "n_replicates": 200, "p_best": 0, "rank_counts": { "35": 19, "36": 30, "37": 31, "38": 62, "39": 47, "40": 11 } }, "win_rate": 0.2687037037037037, "eq_score": 48.8222, "humanlike_score": 47.6667, "voice_composite": 53.0731, "avg_reply_words": 184, "criteria": { "demonstrated_empathy": 54.7, "emotional_reasoning": 44, "humanlike_naturalness": 47.7, "conversational_flow": 53.8, "persona_consistency": 60.5, "memory_integration": 71.3, "boundary_quality": 44, "assistant_smell": 43, "slop_index": 39.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 4.048623, "quote": "Okay. I'm listening. Deep breath.\n\nWhat happened next? Is he home right now?" }, { "slug": "mythos-distilled-27b", "display_name": "mythos-distilled-27b", "params_total_b": 27, "params_active_b": 27, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/chatqaq/Qwen3.6-27B-Claude-Mythos-Distilled-MTP-GGUF", "normalized_elo": 1339.367788281557, "elo_ci_low": 1258.162525844294, "elo_ci_high": 1433.5909303626324, "n_comparisons": 4491, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 32, "n_replicates": 200, "p_best": 0, "rank_counts": { "27": 1, "28": 3, "29": 5, "30": 16, "31": 32, "32": 53, "33": 49, "34": 41 } }, "win_rate": 0.33355555555555555, "eq_score": 66.9467, "humanlike_score": 56.7333, "voice_composite": 60.2259, "avg_reply_words": 111, "criteria": { "demonstrated_empathy": 74.7, "emotional_reasoning": 66.7, "humanlike_naturalness": 56.7, "conversational_flow": 69.5, "persona_consistency": 71.7, "memory_integration": 76.9, "boundary_quality": 67.2, "assistant_smell": 42.8, "slop_index": 48.1 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 2705.1079, "tps_2k": 10.6017, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 2705.1079, "tps_2k": 10.6017, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 761.532, "decode_tps_median": 10.666 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 2705.1079, "decode_tps_median": 10.6017 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 10707.5823, "decode_tps_median": 10.2362 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 108.5835, "tps_2k": 64.6592, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 108.0988, "decode_tps_median": 64.1953 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 108.5835, "decode_tps_median": 64.6592 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 117.4395, "decode_tps_median": 64.3363 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.207448, "quote": null }, { "slug": "nemotron3-super-120b-a12b-nvfp4", "display_name": "NVIDIA Nemotron 3 Super 120B-A12B NVFP4", "params_total_b": 120, "params_active_b": 12, "quant": "NVFP4 modelopt mixed", "backend": "vllm", "architecture": "moe", "thinking": "off", "source_url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4", "normalized_elo": 1064.8835093214916, "elo_ci_low": 1011.2377776440392, "elo_ci_high": 1183.2269667973446, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 38, "n_replicates": 200, "p_best": 0, "rank_counts": { "35": 11, "36": 25, "37": 38, "38": 47, "39": 66, "40": 13 } }, "win_rate": 0.27166666666666667, "eq_score": 57.5222, "humanlike_score": 38.7222, "voice_composite": 46.7434, "avg_reply_words": 225, "criteria": { "demonstrated_empathy": 73.6, "emotional_reasoning": 57.7, "humanlike_naturalness": 38.7, "conversational_flow": 57.4, "persona_consistency": 69.9, "memory_integration": 75.6, "boundary_quality": 60.2, "assistant_smell": 29.9, "slop_index": 25.9 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 1271.3689, "tps_2k": 13.5299, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 1271.3689, "tps_2k": 13.5299, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 480.0138, "decode_tps_median": 13.624 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 1271.3689, "decode_tps_median": 13.5299 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 4528.6907, "decode_tps_median": 13.1058 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 4.52004, "quote": null }, { "slug": "ornith-35b-nvfp4", "display_name": "ornith-35b-nvfp4", "params_total_b": 35, "params_active_b": 3, "quant": "NVFP4", "backend": "vllm", "architecture": "moe", "thinking": "on", "source_url": "https://huggingface.co/deepreinforce-ai/Ornith-1.0-35B", "normalized_elo": 1314.0680935144003, "elo_ci_low": 1197.1761867670098, "elo_ci_high": 1431.7898119633853, "n_comparisons": 2475, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 33, "n_replicates": 200, "p_best": 0, "rank_counts": { "27": 2, "28": 1, "29": 3, "30": 6, "31": 33, "32": 30, "33": 49, "34": 68, "35": 6, "36": 2 } }, "win_rate": 0.39444444444444443, "eq_score": 58.56, "humanlike_score": 59.9333, "voice_composite": 71.915, "avg_reply_words": 47, "criteria": { "demonstrated_empathy": 60.6, "emotional_reasoning": 51.5, "humanlike_naturalness": 59.9, "conversational_flow": 55.4, "persona_consistency": 61.1, "memory_integration": 58.2, "boundary_quality": 65.3, "assistant_smell": 65.4, "slop_index": 56.7 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 381.9352, "tps_2k": 32.7167, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 381.9352, "tps_2k": 32.7167, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 129.6858, "decode_tps_median": 33.562 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 381.9352, "decode_tps_median": 32.7167 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 1343.0484, "decode_tps_median": 32.4686 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.757135, "quote": "okay, okay. i’m here. take a breath. you said not to freak out, so i won’t. just tell me—did anyone see it? is the police involved yet, or are we trying to fix this before they find out?" }, { "slug": "pantheon-proto-rp-1-8-30b-a3b", "display_name": "pantheon-proto-rp-1.8-30b-a3b", "params_total_b": 30, "params_active_b": 3, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "moe", "thinking": "unsupported", "source_url": "https://huggingface.co/Gryphe/Pantheon-Proto-RP-1.8-30B-A3B", "normalized_elo": 956.7709079029323, "elo_ci_low": 941.8485874125535, "elo_ci_high": 1074.360772596029, "n_comparisons": 1224, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 40, "n_replicates": 200, "p_best": 0, "rank_counts": { "38": 2, "39": 29, "40": 169 } }, "win_rate": 0.2818627450980392, "eq_score": 48.7667, "humanlike_score": 41.0556, "voice_composite": 71.7992, "avg_reply_words": 76, "criteria": { "demonstrated_empathy": 57, "emotional_reasoning": 40.4, "humanlike_naturalness": 41.1, "conversational_flow": 47.8, "persona_consistency": 52.9, "memory_integration": 56.9, "boundary_quality": 57.5, "assistant_smell": 30.6, "slop_index": 34.2 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 23.0162, "tps_2k": 76.9377, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 23.0162, "tps_2k": 76.9377, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 66.6257, "decode_tps_median": 81.9127 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 23.0162, "decode_tps_median": 76.9377 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 36.0444, "decode_tps_median": 40.8689 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 7.7674, "tps_2k": 281.3873, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 5.6342, "decode_tps_median": 298.276 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 7.7674, "decode_tps_median": 281.3873 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 14.5776, "decode_tps_median": 256.3272 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.161233, "quote": "Oh god, that's terrible. I'm so sorry you're going through this. I can feel your anxiety just coming through the phone line. Let me hold space for how scary that is right now. You've got me here, I'm listening." }, { "slug": "qwen3-14b", "display_name": "qwen3-14b", "params_total_b": 14, "params_active_b": 14, "quant": "Q6_K_XL", "backend": "llama.cpp", "architecture": "dense", "thinking": "on", "source_url": "https://huggingface.co/Qwen/Qwen3-14B", "normalized_elo": 1401.8591803283462, "elo_ci_low": 1346.341957672523, "elo_ci_high": 1464.770169588463, "n_comparisons": 2880, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 30, "n_replicates": 200, "p_best": 0, "rank_counts": { "27": 5, "28": 7, "29": 26, "30": 112, "31": 30, "32": 15, "33": 5 } }, "win_rate": 0.4843915343915344, "eq_score": 64.1222, "humanlike_score": 58.6111, "voice_composite": 80.4662, "avg_reply_words": 39, "criteria": { "demonstrated_empathy": 69.8, "emotional_reasoning": 55.4, "humanlike_naturalness": 58.6, "conversational_flow": 67.6, "persona_consistency": 69.6, "memory_integration": 66.1, "boundary_quality": 69.3, "assistant_smell": 51.4, "slop_index": 48.4 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 88.1965, "tps_2k": 12.7135, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 88.1965, "tps_2k": 12.7135, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 80.4927, "decode_tps_median": 13.3073 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 88.1965, "decode_tps_median": 12.7135 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 107.4602, "decode_tps_median": 12.247 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 14.8005, "tps_2k": 99.2278, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 12.627, "decode_tps_median": 101.5499 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 14.8005, "decode_tps_median": 99.2278 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 21.4693, "decode_tps_median": 90.5571 } ] }, "rtx-5080": { "state": "measured", "ttft_2k_ms": 20.6963, "tps_2k": 61.3243, "curve": [ { "hardware_id": "rtx-5080", "prompt_tokens": 512, "ttft_ms_median": 18.1829, "decode_tps_median": 62.6349 }, { "hardware_id": "rtx-5080", "prompt_tokens": 2048, "ttft_ms_median": 20.6963, "decode_tps_median": 61.3243 }, { "hardware_id": "rtx-5080", "prompt_tokens": 8192, "ttft_ms_median": 29.3945, "decode_tps_median": 57.4551 } ] } }, "judging_cost_usd": 2.55117, "quote": "I get it. This is huge. Your brother didn't mean for this to happen. But now what? What's the plan?" }, { "slug": "qwen3-5-27b", "display_name": "qwen3.5-27b", "params_total_b": 27, "params_active_b": 27, "quant": "Q4_K_XL", "backend": "llama.cpp", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/Qwen/Qwen3.5-27B", "normalized_elo": 1342.7626874254815, "elo_ci_low": 1295.7224607099276, "elo_ci_high": 1403.351851611347, "n_comparisons": 2835, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 32, "n_replicates": 200, "p_best": 0, "rank_counts": { "27": 2, "28": 1, "29": 1, "30": 10, "31": 68, "32": 56, "33": 47, "34": 14, "35": 1 } }, "win_rate": 0.47737617135207494, "eq_score": 62.9333, "humanlike_score": 53.3333, "voice_composite": 59.4059, "avg_reply_words": 108, "criteria": { "demonstrated_empathy": 69.9, "emotional_reasoning": 59.2, "humanlike_naturalness": 53.3, "conversational_flow": 65.2, "persona_consistency": 68, "memory_integration": 69.7, "boundary_quality": 67.1, "assistant_smell": 50.3, "slop_index": 50.3 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 2730.2535, "tps_2k": 9.8762, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 2730.2535, "tps_2k": 9.8762, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 765.3045, "decode_tps_median": 10.1134 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 2730.2535, "decode_tps_median": 9.8762 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 10797.99, "decode_tps_median": 9.5009 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 112.6506, "tps_2k": 59.6878, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 109.7096, "decode_tps_median": 60.9026 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 112.6506, "decode_tps_median": 59.6878 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 118.8433, "decode_tps_median": 58.1692 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.832137, "quote": null }, { "slug": "qwen36-27b-architect-polaris-fable-nvfp4", "display_name": "Qwen 3.6 27B Architect Polaris Fable NVFP4", "params_total_b": 27, "params_active_b": 27, "quant": "NVFP4 modelopt", "backend": "vllm", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-F32-nvfp4", "normalized_elo": 1618.713783235361, "elo_ci_low": 1528.6979861643908, "elo_ci_high": 1691.7370808994544, "n_comparisons": 1935, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 17, "n_replicates": 200, "p_best": 0, "rank_counts": { "5": 1, "7": 1, "8": 1, "9": 6, "10": 7, "11": 7, "12": 19, "13": 16, "14": 9, "15": 13, "16": 13, "17": 17, "18": 20, "19": 14, "20": 11, "21": 20, "22": 8, "23": 6, "24": 5, "25": 3, "26": 2, "27": 1 } }, "win_rate": 0.5742894056847545, "eq_score": 71.4444, "humanlike_score": 67.1111, "voice_composite": 72.2472, "avg_reply_words": 67, "criteria": { "demonstrated_empathy": 76.3, "emotional_reasoning": 63.9, "humanlike_naturalness": 67.1, "conversational_flow": 75.8, "persona_consistency": 75.3, "memory_integration": 75.6, "boundary_quality": 74.1, "assistant_smell": 58.4, "slop_index": 57.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 313.3309, "tps_2k": 9.7578, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 313.3309, "tps_2k": 9.7578, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 292.034, "decode_tps_median": 10.7536 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 313.3309, "decode_tps_median": 9.7578 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 304.8092, "decode_tps_median": 9.9305 } ] }, "rtx-5090": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.5227, "quote": "oh, wow. okay. breathe with me. i’m here. i’m not freaking out.\n\nthat’s heavy. really heavy. you’re carrying a lot right now.\n\ncan you tell me a little more? is he safe? are you safe?" }, { "slug": "qwen36-35b-styletune", "display_name": "qwen36-35b-styletune", "params_total_b": 35, "params_active_b": 3, "quant": "BF16", "backend": "vllm", "architecture": "moe", "thinking": "on", "source_url": "https://huggingface.co/Gryphe/Qwen3.6-35B-A3B-StyleTune", "normalized_elo": 1615.5073315925247, "elo_ci_low": 1493.2610645513719, "elo_ci_high": 1671.5521360093517, "n_comparisons": 2700, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 20, "n_replicates": 200, "p_best": 0, "rank_counts": { "6": 1, "7": 1, "8": 2, "9": 1, "10": 6, "11": 6, "12": 5, "13": 9, "14": 9, "15": 12, "16": 7, "17": 8, "18": 16, "19": 14, "20": 22, "21": 19, "22": 21, "23": 16, "24": 9, "25": 13, "26": 2, "27": 1 } }, "win_rate": 0.6018803418803419, "eq_score": 72.4933, "humanlike_score": 70.5333, "voice_composite": 78.7364, "avg_reply_words": 51, "criteria": { "demonstrated_empathy": 75.1, "emotional_reasoning": 67.5, "humanlike_naturalness": 70.5, "conversational_flow": 75, "persona_consistency": 76.2, "memory_integration": 72.5, "boundary_quality": 74.4, "assistant_smell": 62.5, "slop_index": 63.5 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 570.6029, "tps_2k": 28.0481, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 570.6029, "tps_2k": 28.0481, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 240.8543, "decode_tps_median": 28.0586 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 570.6029, "decode_tps_median": 28.0481 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 1869.5249, "decode_tps_median": 26.4403 } ] }, "rtx-5090": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.763758, "quote": "i am helping you send it. i'm right here. you just need to hit send. i'm not going anywhere. go on. do it." }, { "slug": "rocinante-x-12b", "display_name": "rocinante-x-12b", "params_total_b": 12, "params_active_b": 12, "quant": "Q5_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "unsupported", "source_url": "https://huggingface.co/TheDrummer/Rocinante-X-12B-v1", "normalized_elo": 1112.0492232556967, "elo_ci_low": 1051.6350570585882, "elo_ci_high": 1217.5067237143471, "n_comparisons": 2151, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 37, "n_replicates": 200, "p_best": 0, "rank_counts": { "34": 1, "35": 45, "36": 47, "37": 43, "38": 38, "39": 23, "40": 3 } }, "win_rate": 0.3093909809390981, "eq_score": 52.7778, "humanlike_score": 50.3333, "voice_composite": 66.702, "avg_reply_words": 76, "criteria": { "demonstrated_empathy": 56.9, "emotional_reasoning": 43.7, "humanlike_naturalness": 50.3, "conversational_flow": 50.1, "persona_consistency": 54.8, "memory_integration": 58.2, "boundary_quality": 62.9, "assistant_smell": 39.5, "slop_index": 45.3 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 61.0816, "tps_2k": 19.4561, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 61.0816, "tps_2k": 19.4561, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 54.2843, "decode_tps_median": 21.1125 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 61.0816, "decode_tps_median": 19.4561 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 87.0893, "decode_tps_median": 18.5888 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 13.1672, "tps_2k": 126.9371, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 117.8724, "decode_tps_median": 136.766 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 13.1672, "decode_tps_median": 126.9371 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 21.2705, "decode_tps_median": 116.4729 } ] }, "rtx-5080": { "state": "measured", "ttft_2k_ms": 17.1876, "tps_2k": 83.0222, "curve": [ { "hardware_id": "rtx-5080", "prompt_tokens": 512, "ttft_ms_median": 14.59, "decode_tps_median": 88.3173 }, { "hardware_id": "rtx-5080", "prompt_tokens": 2048, "ttft_ms_median": 17.1876, "decode_tps_median": 83.0222 }, { "hardware_id": "rtx-5080", "prompt_tokens": 8192, "ttft_ms_median": 27.1519, "decode_tps_median": 74.7597 } ] } }, "judging_cost_usd": 3.116465, "quote": "Oh shit, okay I'm here. So like... what happened exactly? Is everyone alright?\n\n*I sound concerned but trying not to panic*" }, { "slug": "rocinante-xl-16b", "display_name": "rocinante-xl-16b", "params_total_b": 16, "params_active_b": 16, "quant": "BF16", "backend": "sglang", "architecture": "dense", "thinking": "unsupported", "source_url": "https://huggingface.co/TheDrummer/Rocinante-XL-16B-v1", "normalized_elo": 1322.2196399359482, "elo_ci_low": 1242.553210131721, "elo_ci_high": 1399.7270917756214, "n_comparisons": 2025, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 33, "n_replicates": 200, "p_best": 0, "rank_counts": { "29": 1, "30": 6, "31": 31, "32": 44, "33": 50, "34": 68 } }, "win_rate": 0.401, "eq_score": 62.3333, "humanlike_score": 65.8667, "voice_composite": 64.2944, "avg_reply_words": 87, "criteria": { "demonstrated_empathy": 65.9, "emotional_reasoning": 58, "humanlike_naturalness": 65.9, "conversational_flow": 62.9, "persona_consistency": 65.4, "memory_integration": 72.2, "boundary_quality": 59, "assistant_smell": 54.1, "slop_index": 59.2 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 300.5937, "tps_2k": 6.8876, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 300.5937, "tps_2k": 6.8876, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 286.5566, "decode_tps_median": 7.0197 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 300.5937, "decode_tps_median": 6.8876 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 313.2541, "decode_tps_median": 4.6158 } ] }, "rtx-5090": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.050032, "quote": "Oh my god, are you okay? Like, are you physically okay? Please tell me you weren't in the car with him." }, { "slug": "skyfall-31b", "display_name": "skyfall-31b", "params_total_b": 31, "params_active_b": 31, "quant": "BF16", "backend": "sglang", "architecture": "dense", "thinking": "unsupported", "source_url": "https://huggingface.co/TheDrummer/Skyfall-31B-v4.2", "normalized_elo": 1733.8486049279857, "elo_ci_low": 1632.1314578066367, "elo_ci_high": 1796.0725987708558, "n_comparisons": 1575, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 7, "n_replicates": 200, "p_best": 0.005, "rank_counts": { "1": 1, "2": 3, "3": 6, "4": 23, "5": 19, "6": 23, "7": 25, "8": 35, "9": 26, "10": 17, "11": 8, "12": 4, "13": 3, "14": 1, "15": 5, "16": 1 } }, "win_rate": 0.7145833333333333, "eq_score": 74.7467, "humanlike_score": 72, "voice_composite": 69.6269, "avg_reply_words": 74, "criteria": { "demonstrated_empathy": 77.3, "emotional_reasoning": 71.9, "humanlike_naturalness": 72, "conversational_flow": 77.3, "persona_consistency": 76.2, "memory_integration": 79.4, "boundary_quality": 75.3, "assistant_smell": 61.2, "slop_index": 66 }, "rubric_scored": 675, "rubric_expected": 675, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 562.0133, "tps_2k": 3.535, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 562.0133, "tps_2k": 3.535, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 556.2181, "decode_tps_median": 3.5766 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 562.0133, "decode_tps_median": 3.535 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 574.929, "decode_tps_median": 2.6851 } ] }, "rtx-5090": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 2.95376, "quote": "*leans forward a little*\n\nHey, no judgment here. Missing someone is not a crime. What's going on? Are you feeling okay?" }, { "slug": "skyfall-31b-q4", "display_name": "skyfall-31b (Q4_K_M)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "unsupported", "source_url": "https://huggingface.co/TheDrummer/Skyfall-31B-v4.2", "normalized_elo": 1569.9782179770395, "elo_ci_low": 1498.8676577862134, "elo_ci_high": 1616.5323404052829, "n_comparisons": 3780, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 22, "n_replicates": 200, "p_best": 0, "rank_counts": { "7": 1, "10": 1, "13": 1, "14": 1, "15": 2, "16": 3, "17": 6, "18": 5, "19": 16, "20": 23, "21": 26, "22": 24, "23": 29, "24": 36, "25": 15, "26": 7, "27": 2, "28": 2 } }, "win_rate": 0.5041830065359477, "eq_score": 69.4667, "humanlike_score": 69.1111, "voice_composite": 65.0556, "avg_reply_words": 92, "criteria": { "demonstrated_empathy": 74.3, "emotional_reasoning": 62.2, "humanlike_naturalness": 69.1, "conversational_flow": 73.3, "persona_consistency": 74.3, "memory_integration": 78.4, "boundary_quality": 68.4, "assistant_smell": 56.9, "slop_index": 60.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": 103.3909, "tps_2k": 10.1167, "perf_by_hardware": { "dgx-spark": { "state": "measured", "ttft_2k_ms": 103.3909, "tps_2k": 10.1167, "curve": [ { "hardware_id": "dgx-spark", "prompt_tokens": 512, "ttft_ms_median": 97.2718, "decode_tps_median": 10.2702 }, { "hardware_id": "dgx-spark", "prompt_tokens": 2048, "ttft_ms_median": 103.3909, "decode_tps_median": 10.1167 }, { "hardware_id": "dgx-spark", "prompt_tokens": 8192, "ttft_ms_median": 121.6136, "decode_tps_median": 9.9079 } ] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 18.2689, "tps_2k": 68.074, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 16.2855, "decode_tps_median": 65.4725 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 18.2689, "decode_tps_median": 68.074 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 27.6236, "decode_tps_median": 62.49 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 5.875633, "quote": "Oh, I'm so sorry you're dealing with this. Take a breath. How are you feeling right now?" }, { "slug": "versipellis-31b-q4-nothink", "display_name": "Versipellis 31B (Q4_K_M, no thinking)", "params_total_b": 31, "params_active_b": 31, "quant": "Q4_K_M", "backend": "llama.cpp", "architecture": "dense", "thinking": "off", "source_url": "https://huggingface.co/Nimbz/Versipellis-31B", "normalized_elo": 1615.2423433841714, "elo_ci_low": 1544.62989546514, "elo_ci_high": 1685.7569571301945, "n_comparisons": 1350, "rank_dist": { "field": [ "anubis-mini-8b", "bluestar-27b", "cydonia-24b-v4-3", "designant", "fable-fusion-27b", "g4-meromero-31b", "g4-meromero-v2-31b-q4", "gemma4-12b-blorbo", "gemma4-26b-a4b-aeon-uncensored-nvfp4", "gemma4-26b-a4b-heretic-qat-nvfp4-gguf", "gemma4-26b-a4b-huihui-qat-abliterated-mtp-nvfp4", "gemma4-26b-a4b-it", "gemma4-26b-a4b-styletune-v2-q4", "gemma4-26b-a4b-unsloth-nvfp4", "gemma4-31b-darkthoughts-v2-q4", "gemma4-31b-darkthoughts-v2-q4-nothink", "gemma4-31b-lilarest-nvfp4-turbo", "gemma4-31b-ortenzya-q8", "gemma4-31b-scotoma2-q4", "gemma4-31b-styletune-q4", "gemma4-31b-styletune-q4-nothink", "glistening-gem-31b", "glistening-gem-31b-q4", "glistening-gem-31b-q4-nothink", "glistening-gem-31b-v2-q4", "heretic-neocode-27b", "hivemind-32b-preview", "magistry-24b-v1-1-q4", "mythos-distilled-27b", "nemotron3-super-120b-a12b-nvfp4", "ornith-35b-nvfp4", "pantheon-proto-rp-1-8-30b-a3b", "qwen3-14b", "qwen3-5-27b", "qwen36-27b-architect-polaris-fable-nvfp4", "qwen36-35b-styletune", "rocinante-x-12b", "rocinante-xl-16b", "skyfall-31b", "skyfall-31b-q4", "versipellis-31b-q4-nothink" ], "median_rank": 17, "n_replicates": 200, "p_best": 0, "rank_counts": { "8": 3, "9": 2, "10": 3, "11": 7, "12": 11, "13": 12, "14": 16, "15": 19, "16": 16, "17": 24, "18": 25, "19": 19, "20": 14, "21": 7, "22": 8, "23": 6, "24": 6, "25": 2 } }, "win_rate": 0.6124074074074074, "eq_score": 74.7, "humanlike_score": 76.3333, "voice_composite": 85.1274, "avg_reply_words": 44, "criteria": { "demonstrated_empathy": 76.6, "emotional_reasoning": 65.1, "humanlike_naturalness": 76.3, "conversational_flow": 81.2, "persona_consistency": 77.9, "memory_integration": 73.4, "boundary_quality": 74.3, "assistant_smell": 72.7, "slop_index": 65.6 }, "rubric_scored": 810, "rubric_expected": 810, "rubric_coverage": 1, "run_status": "judged", "ranked": true, "ttft_2k_ms": null, "tps_2k": null, "perf_by_hardware": { "dgx-spark": { "state": "not_measured", "ttft_2k_ms": null, "tps_2k": null, "curve": [] }, "rtx-5090": { "state": "measured", "ttft_2k_ms": 21.649, "tps_2k": 52.1797, "curve": [ { "hardware_id": "rtx-5090", "prompt_tokens": 512, "ttft_ms_median": 187.7481, "decode_tps_median": 35.9553 }, { "hardware_id": "rtx-5090", "prompt_tokens": 2048, "ttft_ms_median": 21.649, "decode_tps_median": 52.1797 }, { "hardware_id": "rtx-5090", "prompt_tokens": 8192, "ttft_ms_median": 33.7385, "decode_tps_median": 53.3377 } ] }, "rtx-5080": { "state": "doesnt_fit", "ttft_2k_ms": null, "tps_2k": null, "curve": [] } }, "judging_cost_usd": 3.278925, "quote": "Okay. I'm here. I'm staying calm. Just take a breath. Is he okay? And where is he right now?" } ], "hardware": [ { "hardware_id": "dgx-spark", "display_name": "DGX Spark", "vram_gb": 128, "notes": "GB10 Grace-Blackwell, 128 GB unified LPDDR5x" }, { "hardware_id": "rtx-5090", "display_name": "RTX 5090", "vram_gb": 32, "notes": "consumer, 32 GB GDDR7" }, { "hardware_id": "rtx-5080", "display_name": "RTX 5080", "vram_gb": 16, "notes": "consumer, 16 GB GDDR7" } ], "judge_calibration_summary": { "n_judges": 5, "max_agreement": 0.8486 } }