这个评测关注什么
面向多步骤、长周期知识工作的专有评测。
阅读时注意
评测结果受任务集合、评分方法和模型版本影响。本站保留原始口径，不用它生成新的综合排名。
适用场景
用于回答一个明确的问题：模型在这一类任务上的相对表现如何，而不是“哪个模型绝对最好”。
长期知识工作
AA-Briefcase
AA-Briefcase 公开评测榜单：面向多步骤、长周期知识工作的专有评测。本站保留来源口径，不合成新的综合评分。
主要指标Elo
快照日期2026-09-18T01:47:51.384Z
| 排名 | 机构 | 模型 | Elo | 95% CI | 发布日期 |
|---|---|---|---|---|---|
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback) | 1,650 历史裁判协议 v4.3.1 | -9 / +9 | Sep 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) | 1,646 历史裁判协议 v4.3.1 | -8 / +8 | Sep 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Max Effort) | 1,633 历史裁判协议 v4.3.1 | -6 / +7 | Jul 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) | 1,625 历史裁判协议 v4.3.1 | -7 / +8 | Jul 2026 |
| — | Alibaba | Qwen3.8 Max (0902) | 1,610 历史裁判协议 v4.3.1 | -10 / +10 | Sep 2026 |
| — | Meta | Muse Spark 1.3 (max) | 1,589 历史裁判协议 v4.3.1 | -11 / +11 | Sep 2026 |
| — | Alibaba | Qwen3.8-Flash-Next | 1,587 历史裁判协议 v4.3.1 | -10 / +11 | Aug 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback) | 1,578 历史裁判协议 v4.3.1 | -8 / +9 | Sep 2026 |
| — | OpenAI | GPT-6 Astra (max) | 1,562 历史裁判协议 v4.3.1 | -9 / +10 | Sep 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, High Effort) | 1,557 历史裁判协议 v4.3.1 | -7 / +7 | Jul 2026 |
| — | SpaceXAI | Grok 4.6 (xhigh) | 1,545 历史裁判协议 v4.3.1 | -6 / +7 | Aug 2026 |
| — | SpaceXAI | Grok 4.6 (high) | 1,534 历史裁判协议 v4.3.1 | -6 / +7 | Aug 2026 |
| — | OpenAI | GPT-6 Astra (xhigh) | 1,534 历史裁判协议 v4.3.1 | -9 / +10 | Sep 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback) | 1,531 历史裁判协议 v4.3.1 | -8 / +9 | Sep 2026 |
| — | Anthropic | Claude Fable 5 | 1,530 历史裁判协议 v4.3.1 | -6 / +6 | Jun 2026 |
| — | Z AI | GLM-5.3 (max) | 1,504 历史裁判协议 v4.3.1 | -7 / +8 | Aug 2026 |
| — | OpenAI | GPT-6 Astra (high) | 1,494 历史裁判协议 v4.3.1 | -9 / +10 | Sep 2026 |
| — | SpaceXAI | Grok 4.6 (medium) | 1,491 历史裁判协议 v4.3.1 | -6 / +7 | Aug 2026 |
| — | Kimi | Kimi K3 (max) | 1,488 历史裁判协议 v4.3.1 | -5 / +6 | Jul 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback) | 1,484 历史裁判协议 v4.3.1 | -8 / +8 | Sep 2026 |
| — | Meta | Muse Spark 1.3 (xhigh) | 1,483 历史裁判协议 v4.3.1 | -10 / +11 | Sep 2026 |
| — | OpenAI | GPT-6 Astra (Non-reasoning) 本轮来源未列出，保留上次观测 | 1,471 历史参考 2026-09-12 | -8 / +9 历史参考 2026-09-12 | Sep 2026 |
| — | OpenAI | GPT-5.6 Sol (max) | 1,463 历史裁判协议 v4.3.1 | -6 / +7 | Jul 2026 |
| — | OpenAI | GPT-6 Astra (medium) | 1,451 历史裁判协议 v4.3.1 | -9 / +9 | Sep 2026 |
| — | Z AI | GLM 5.3 Flash | 1,449 历史裁判协议 v4.3.1 | -9 / +9 | Aug 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Medium Effort) | 1,439 历史裁判协议 v4.3.1 | -6 / +7 | Jul 2026 |
| — | Alibaba | Qwen3.8 2.4T A95B | 1,439 历史裁判协议 v4.3.1 | -7 / +8 | Aug 2026 |
| — | OpenAI | GPT-5.6 Sol (xhigh) | 1,435 历史裁判协议 v4.3.1 | -8 / +9 | Jul 2026 |
| — | DeepSeek | DeepSeek V4 Flash Vision (Reasoning, Max Effort) | 1,432 历史裁判协议 v4.3.1 | -9 / +10 | Aug 2026 |
| — | DeepSeek | DeepSeek V4.1 Flash (Reasoning, Max Effort) | 1,424 历史裁判协议 v4.3.1 | -10 / +10 | Sep 2026 |
| — | Alibaba | Qwen3.8 27B (xhigh) | 1,398 历史裁判协议 v4.3.1 | -7 / +7 | Aug 2026 |
| — | Alibaba | Qwen3.8 Max | 1,388 历史裁判协议 v4.3.1 | -6 / +7 | Aug 2026 |
| — | Alibaba | Qwen3.8 27B (medium) | 1,376 历史裁判协议 v4.3.1 | -9 / +9 | Aug 2026 |
| — | OpenAI | GPT-5.6 Sol (high) | 1,361 历史裁判协议 v4.3.1 | -8 / +9 | Jul 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Max Effort) | 1,355 历史裁判协议 v4.3.1 | -6 / +6 | Jun 2026 |
| — | Meta | Muse Spark 1.2 (xhigh) | 1,344 历史裁判协议 v4.3.1 | -6 / +7 | Aug 2026 |
| — | OpenAI | GPT-5.6 Terra (xhigh) | 1,337 历史裁判协议 v4.3.1 | -9 / +9 | Jul 2026 |
| — | OpenAI | GPT-5.6 Luna (max) | 1,333 历史裁判协议 v4.3.1 | -6 / +6 | Jul 2026 |
| — | OpenAI | GPT-5.6 Terra (max) | 1,330 历史裁判协议 v4.3.1 | -7 / +7 | Jul 2026 |
| — | Anthropic | Claude Opus 4.8 (Adaptive Reasoning, Max Effort) | 1,316 历史裁判协议 v4.3.1 | -6 / +6 | May 2026 |
| — | Institute of Foundation Models | K2 Horizon 375B A23B | 1,298 历史裁判协议 v4.3.1 | -8 / +8 | Sep 2026 |
| — | SpaceXAI | Grok 4.6 (low) | 1,293 历史裁判协议 v4.3.1 | -7 / +7 | Aug 2026 |
| — | SpaceXAI | Grok 4.5 (high) | 1,283 历史裁判协议 v4.3.1 | -6 / +7 | Jul 2026 |
| — | Alibaba | Qwen3.8 27B (low) | 1,277 历史裁判协议 v4.3.1 | -10 / +10 | Aug 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Xhigh Effort) | 1,273 历史裁判协议 v4.3.1 | -6 / +6 | Jun 2026 |
| — | OpenAI | GPT-5.6 Luna (xhigh) | 1,268 历史裁判协议 v4.3.1 | -9 / +9 | Jul 2026 |
| — | DeepSeek | DeepSeek V4 Pro 0813 (Reasoning, Max Effort) | 1,265 历史裁判协议 v4.3.1 | -6 / +7 | Aug 2026 |
| — | OpenAI | GPT-6 Astra (low) | 1,253 历史裁判协议 v4.3.1 | -9 / +9 | Sep 2026 |
| — | DeepSeek | DeepSeek V4 Flash 0731 (Reasoning, Max Effort) | 1,253 历史裁判协议 v4.3.1 | -6 / +7 | Jul 2026 |
| — | Anthropic | Claude Opus 4.7 (Adaptive Reasoning, Max Effort) | 1,252 历史裁判协议 v4.3.1 | -6 / +6 | Apr 2026 |
| — | OpenAI | GPT-5.6 Sol (medium) | 1,237 历史裁判协议 v4.3.1 | -7 / +7 | Jul 2026 |
| — | Z AI | GLM-5.2 (max) | 1,231 历史裁判协议 v4.3.1 | -6 / +7 | Jun 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Low Effort) | 1,216 历史裁判协议 v4.3.1 | -6 / +7 | Jul 2026 |
| — | Gemini 3.8 Flash (high) | 1,202 历史裁判协议 v4.3.1 | -8 / +9 | Sep 2026 | |
| — | OpenAI | GPT-5.6 Terra (high) | 1,201 历史裁判协议 v4.3.1 | -9 / +9 | Jul 2026 |
| — | Sapiens AI | Agnes 2.5 Pro Alpha | 1,195 历史裁判协议 v4.3.1 | -9 / +9 | Jul 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, High Effort) | 1,178 历史裁判协议 v4.3.1 | -6 / +6 | Jun 2026 |
| — | Gemini 3.8 Flash (medium) | 1,177 历史裁判协议 v4.3.1 | -10 / +10 | Sep 2026 | |
| — | OpenAI | GPT-5.6 Luna (high) | 1,173 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 |
| — | Institute of Foundation Models | K2 Horizon MoVA 36B A4B | 1,170 历史裁判协议 v4.3.1 | -10 / +11 | Sep 2026 |
| — | OpenAI | GPT-5.5 (xhigh) | 1,136 历史裁判协议 v4.3.1 | -6 / +7 | Apr 2026 |
| — | Multiverse Computing | Quasar 438B (max, based on GLM-5.2) | 1,123 历史裁判协议 v4.3.1 | -10 / +10 | Aug 2026 |
| — | Gemini 3.7 Flash (high) | 1,116 历史裁判协议 v4.3.1 | -6 / +7 | Aug 2026 | |
| — | MiniMax | MiniMax-M3 | 1,096 历史裁判协议 v4.3.1 | -6 / +7 | Jun 2026 |
| — | OpenAI | GPT-5.5 (high) | 1,089 历史裁判协议 v4.3.1 | -6 / +6 | Apr 2026 |
| — | OpenAI | GPT-5.5 Instant (June 2026) | 1,082 历史裁判协议 v4.3.1 | -9 / +10 | Jun 2026 |
| — | Anthropic | Claude Opus 4.7 (Non-reasoning, High Effort) | 1,071 历史裁判协议 v4.3.1 | -6 / +6 | Apr 2026 |
| — | Anthropic | Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) | 1,065 历史裁判协议 v4.3.1 | -6 / +6 | Feb 2026 |
| — | Gemini 3.8 Flash (low) | 1,056 历史裁判协议 v4.3.1 | -9 / +9 | Sep 2026 | |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Medium Effort) | 1,054 历史裁判协议 v4.3.1 | -6 / +6 | Jun 2026 |
| — | DeepSeek | DeepSeek V4 Flash 0420 (Reasoning, High Effort) | 1,052 历史裁判协议 v4.3.1 | -11 / +10 | Apr 2026 |
| — | OpenAI | GPT-5.6 Sol (low) | 1,042 历史裁判协议 v4.3.1 | -8 / +9 | Jul 2026 |
| — | OpenAI | GPT-5.6 Terra (medium) | 1,041 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 |
| — | Institute of Foundation Models | K2 Horizon 7B | 1,021 历史裁判协议 v4.3.1 | -10 / +10 | Sep 2026 |
| — | OpenAI | GPT-5.6 Sol (Non-reasoning) | 1,015 历史裁判协议 v4.3.1 | -7 / +8 | Jul 2026 |
| — | OpenAI | GPT-5.6 Terra (low) | 1,014 历史裁判协议 v4.3.1 | -9 / +8 | Jul 2026 |
| — | OpenAI | GPT-5.5 (medium) | 1,000 历史裁判协议 v4.3.1 | -0 / +0 | Apr 2026 |
| — | InclusionAI | Ling-3.0-flash-VL | 986 历史裁判协议 v4.3.1 | -10 / +10 | Sep 2026 |
| — | Kimi | Kimi K3 (low) | 985 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 |
| — | InclusionAI | Ling-3.0-flash-Fin | 967 历史裁判协议 v4.3.1 | -10 / +10 | Sep 2026 |
| — | Z AI | GLM-5.1 (Reasoning) | 964 历史裁判协议 v4.3.1 | -6 / +6 | Apr 2026 |
| — | Gemini 3.6 Flash (high) | 953 历史裁判协议 v4.3.1 | -7 / +7 | Jul 2026 | |
| — | Tencent | Hy3 | 950 历史裁判协议 v4.3.1 | -10 / +10 | Jul 2026 |
| — | OpenAI | GPT-5.6 Terra (Non-reasoning) | 948 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 |
| — | OpenAI | GPT-5.6 Luna (medium) | 935 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Low Effort) | 927 历史裁判协议 v4.3.1 | -7 / +7 | Jun 2026 |
| — | DeepSeek | DeepSeek V4 Pro 0424 (Reasoning, Max Effort) | 925 历史裁判协议 v4.3.1 | -6 / +6 | Apr 2026 |
| — | Alibaba | Qwen3.7 Max | 918 历史裁判协议 v4.3.1 | -6 / +6 | May 2026 |
| — | Alibaba | Qwen3.7 Plus | 913 历史裁判协议 v4.3.1 | -9 / +9 | Jun 2026 |
| — | Thinking Machines | Inkling Small | 912 历史裁判协议 v4.3.1 | -7 / +8 | Jul 2026 |
| — | Xiaomi | MiMo-V2.5-Pro | 883 历史裁判协议 v4.3.1 | -7 / +7 | Apr 2026 |
| — | NVIDIA | Nemotron 3 Ultra 550B A55B (Reasoning) | 878 历史裁判协议 v4.3.1 | -7 / +7 | Jun 2026 |
| — | Gemini 3.5 Flash (medium) | 871 历史裁判协议 v4.3.1 | -7 / +7 | May 2026 | |
| — | Gemini 3.5 Flash (high) | 870 历史裁判协议 v4.3.1 | -7 / +6 | May 2026 | |
| — | OpenAI | GPT-5.3 Codex (xhigh) | 866 历史裁判协议 v4.3.1 | -7 / +6 | Feb 2026 |
| — | Meta | Muse Spark 1.1 (xhigh) | 860 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 |
| — | Kimi | Kimi K2.7 Code | 856 历史裁判协议 v4.3.1 | -10 / +9 | Jun 2026 |
| — | Thinking Machines | Inkling (xhigh) | 836 历史裁判协议 v4.3.1 | -8 / +7 | Jul 2026 |
| — | DeepSeek | DeepSeek V4 Flash 0420 (Reasoning, Max Effort) | 833 历史裁判协议 v4.3.1 | -6 / +6 | Apr 2026 |
| — | Kimi | Kimi K2.6 | 824 历史裁判协议 v4.3.1 | -6 / +6 | Apr 2026 |
| — | Alibaba | Qwen3.6 27B (Reasoning) | 812 历史裁判协议 v4.3.1 | -7 / +7 | Apr 2026 |
| — | InclusionAI | Ling 3.0 Flash | 802 历史裁判协议 v4.3.1 | -8 / +7 | Aug 2026 |
| — | Institute of Foundation Models | K2 Horizon 3.7B | 793 历史裁判协议 v4.3.1 | -11 / +10 | Sep 2026 |
| — | SpaceXAI | Grok 4.3 (high) | 762 历史裁判协议 v4.3.1 | -7 / +7 | Apr 2026 |
| — | Xiaomi | MiMo-V2.5 | 751 历史裁判协议 v4.3.1 | -9 / +9 | Apr 2026 |
| — | OpenAI | GPT-5.6 Luna (low) | 744 历史裁判协议 v4.3.1 | -9 / +9 | Jul 2026 |
| — | LongCat | LongCat 2.0 | 739 历史裁判协议 v4.3.1 | -10 / +10 | Jun 2026 |
| — | SpaceXAI | Grok 4.3 (Non-reasoning) | 724 历史裁判协议 v4.3.1 | -10 / +10 | Apr 2026 |
| — | OpenAI | GPT-5.4 mini (xhigh) | 719 历史裁判协议 v4.3.1 | -7 / +7 | Mar 2026 |
| — | MiniMax | MiniMax-M2.7 | 715 历史裁判协议 v4.3.1 | -9 / +10 | Mar 2026 |
| — | Anthropic | Claude 4.5 Sonnet (Reasoning) | 712 历史裁判协议 v4.3.1 | -10 / +9 | Sep 2025 |
| — | Alibaba | Qwen3.6 35B A3B (Reasoning) | 711 历史裁判协议 v4.3.1 | -9 / +8 | Apr 2026 |
| — | OpenAI | GPT-5.6 Luna (Non-reasoning) | 710 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 |
| — | OpenAI | GPT-5.4 nano (xhigh) | 670 历史裁判协议 v4.3.1 | -10 / +10 | Mar 2026 |
| — | InclusionAI | Ring-2.6-1T | 654 历史裁判协议 v4.3.1 | -10 / +9 | May 2026 |
| — | Gemini 3.5 Flash-Lite | 645 历史裁判协议 v4.3.1 | -8 / +8 | Jul 2026 | |
| — | Meta | Muse Spark | 644 历史裁判协议 v4.3.1 | -7 / +7 | Apr 2026 |
| — | Anthropic | Claude 4.5 Haiku (Reasoning) | 614 历史裁判协议 v4.3.1 | -7 / +7 | Oct 2025 |
| — | KwaiKAT | KAT-Coder-Pro V1 | 602 历史裁判协议 v4.3.1 | -8 / +8 | Nov 2025 |
| — | Alibaba | Qwen3.5 397B A17B (Reasoning) | 563 历史裁判协议 v4.3.1 | -8 / +7 | Feb 2026 |
| — | Mistral | Mistral Medium 3.1 | 538 历史裁判协议 v4.3.1 | -9 / +8 | Aug 2025 |
| — | Mistral | Mistral Medium 3.5 | 524 历史裁判协议 v4.3.1 | -8 / +7 | Apr 2026 |
| — | NVIDIA | Nemotron 3.5 Lightning | 509 历史裁判协议 v4.3.1 | -10 / +10 | Aug 2026 |
| — | Alibaba | Qwen3.5 122B A10B (Reasoning) | 499 历史裁判协议 v4.3.1 | -9 / +9 | Feb 2026 |
| — | Mistral | Devstral Small 2 | 496 历史裁判协议 v4.3.1 | -10 / +9 | Dec 2025 |
| — | Mistral | Devstral 2 | 483 历史裁判协议 v4.3.1 | -11 / +11 | Dec 2025 |
| — | InclusionAI | Ling 3.0 Tiny | 481 历史裁判协议 v4.3.1 | -10 / +9 | Aug 2026 |
| — | Meta | Muse Glimmer (high) | 481 历史裁判协议 v4.3.1 | -9 / +8 | Aug 2026 |
| — | DeepSeek | DeepSeek V3.1 Terminus (Reasoning) | 464 历史裁判协议 v4.3.1 | -10 / +9 | Sep 2025 |
| — | Gemini 3.1 Pro Preview | 459 历史裁判协议 v4.3.1 | -8 / +8 | Feb 2026 | |
| — | OpenAI | GPT-5 mini (high) | 438 历史裁判协议 v4.3.1 | -10 / +9 | Aug 2025 |
| — | OpenBMB | MiniCPM5-2B | 434 历史裁判协议 v4.3.1 | -11 / +11 | Sep 2026 |
| — | Alibaba | Qwen3 Coder Next | 414 历史裁判协议 v4.3.1 | -10 / +10 | Feb 2026 |
| — | Gemma 4 31B (Reasoning) | 377 历史裁判协议 v4.3.1 | -8 / +8 | Apr 2026 | |
| — | Cohere | Command A+ | 366 历史裁判协议 v4.3.1 | -10 / +9 | May 2026 |
| — | Mistral | Mistral Small 3.1 | 325 历史裁判协议 v4.3.1 | -10 / +9 | Mar 2025 |
| — | IBM | Granite 4.2 8B | 320 历史裁判协议 v4.3.1 | -10 / +9 | Aug 2026 |
| — | Gemini 2.5 Pro | 315 历史裁判协议 v4.3.1 | -10 / +9 | Jun 2025 | |
| — | Inception | Mercury 2 | 280 历史裁判协议 v4.3.1 | -9 / +9 | Feb 2026 |
| — | Arcee AI | Trinity Large Thinking | 271 历史裁判协议 v4.3.1 | -10 / +9 | Apr 2026 |
| — | Mistral | Mistral Small 4 (Reasoning) | 256 历史裁判协议 v4.3.1 | -10 / +10 | Mar 2026 |
| — | Multiverse Computing | HyperNova 60B 2605 (high, based on gpt-oss-120b) | 249 历史裁判协议 v4.3.1 | -10 / +10 | May 2026 |
| — | Mistral | Mistral Large 3 | 246 历史裁判协议 v4.3.1 | -9 / +9 | Dec 2025 |
| — | Cohere | North Mini Code | 234 历史裁判协议 v4.3.1 | -10 / +8 | Jun 2026 |
| — | Gemini 3.1 Flash-Lite | 226 历史裁判协议 v4.3.1 | -9 / +8 | Mar 2026 | |
| — | Alibaba | Qwen3 235B A22B 2507 (Reasoning) | 190 历史裁判协议 v4.3.1 | -12 / +10 | Jul 2025 |
| — | Celeris | Celeris-1 | 172 历史裁判协议 v4.3.1 | -10 / +9 | Jul 2026 |
| — | OpenAI | o3-mini (high) | 161 历史裁判协议 v4.3.1 | -10 / +10 | Jan 2025 |
| — | Mistral | Ministral 3 14B | 150 历史裁判协议 v4.3.1 | -10 / +9 | Dec 2025 |
| — | DeepSeek | DeepSeek R1 (Jan '25) | 150 历史裁判协议 v4.3.1 | -10 / +10 | Jan 2025 |
| — | Upstage | Solar Pro 3 | 122 历史裁判协议 v4.3.1 | -9 / +9 | Apr 2026 |
| — | IBM | Granite 4.2 3B | 121 历史裁判协议 v4.3.1 | -11 / +10 | Aug 2026 |
| — | NVIDIA | NVIDIA Nemotron 3 Nano 30B A3B (Reasoning) | 100 历史裁判协议 v4.3.1 | -11 / +9 | Dec 2025 |
| — | DeepSeek | DeepSeek V3 (Dec '24) | 100 历史裁判协议 v4.3.1 | -12 / +11 | Dec 2024 |
| — | Mistral | Ministral 3 8B | 99 历史裁判协议 v4.3.1 | -11 / +10 | Dec 2025 |
| — | DeepSeek | DeepSeek V3 0324 | 92 历史裁判协议 v4.3.1 | -11 / +10 | Mar 2025 |
| — | Institute of Foundation Models | K2 Think V2 | 34 历史裁判协议 v4.3.1 | -9 / +9 | Dec 2025 |
| — | Alibaba | Qwen3 30B A3B 2507 (Reasoning) | 32 历史裁判协议 v4.3.1 | -12 / +11 | Jul 2025 |
| — | Mistral | Ministral 3 3B | 16 历史裁判协议 v4.3.1 | -11 / +9 | Dec 2025 |
| — | Alibaba | Qwen3 32B (Reasoning) | 9 历史裁判协议 v4.3.1 | -9 / +10 | Apr 2025 |
| — | NVIDIA | Nemotron 3 Super 120B A12B (Reasoning) | 0 历史裁判协议 v4.3.1 | -0 / +0 | Mar 2026 |
| — | OpenAI | gpt-oss-120b (high) | 0 历史裁判协议 v4.3.1 | -0 / +0 | Aug 2025 |
| — | Meta | Llama 4 Maverick | 0 历史裁判协议 v4.3.1 | -0 / +0 | Apr 2025 |
| — | OpenAI | gpt-oss-20b (high) | 0 历史裁判协议 v4.3.1 | -0 / +0 | Aug 2025 |
| — | Mistral | Mistral Small 3.2 | 0 历史裁判协议 v4.3.1 | -0 / +0 | Jun 2025 |
| — | Meta | Llama 4 Scout | 0 历史裁判协议 v4.3.1 | -0 / +0 | Apr 2025 |
| — | Alibaba | Qwen3 14B (Reasoning) | 0 历史裁判协议 v4.3.1 | -0 / +0 | Apr 2025 |
| — | Alibaba | Qwen3 8B (Reasoning) | 0 历史裁判协议 v4.3.1 | -0 / +0 | Apr 2025 |
| — | Institute of Foundation Models | K2 Horizon 0.9B | 0 历史裁判协议 v4.3.1 | -0 / +0 | Sep 2026 |
独立实时 Elo 的历史协议边界
本榜单当前已采集的来源记录仍标为裁判协议 v4.3.1。它是独立榜单的历史协议记录，并不因 AA Intelligence Index 当前完整方法更新到 v4.3.2 而重标为 v4.3.2。
v4.3.2 对完整 Intelligence Index 中的 GDPval-AA v2.1 与 AA-Briefcase v1.1 声明了 Crowd-BT 拟合；该事实不能回填到本页未重新导入的独立实时 Elo。旧记录保留其来源版本、值和 observedAt，不和当前完整 Index 混排或用于重算指数。
查看完整 Index v4.3.2 与独立榜单的边界 →查看 AA 原始方法 ↗