这个评测关注什么
覆盖真实职业任务和高经济价值知识工作。
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”。
经济价值任务
GDPval-AA v2
GDPval-AA v2 公开评测榜单：覆盖真实职业任务和高经济价值知识工作。本站保留来源口径，不合成新的综合评分。
主要指标Leaderboard
快照日期2026-09-18T01:47:51.386Z
| 排名 | 机构 | 模型 | Elo | 95% CI | 发布日期 |
|---|---|---|---|---|---|
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback) | 1,745 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Max Effort) | 1,735 历史裁判协议 v4.3.1 | -16 / +16 | Jul 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback) | 1,724 历史裁判协议 v4.3.1 | -16 / +16 | Sep 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Xhigh Effort) | 1,708 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | Meta | Muse Spark 1.3 (max) | 1,703 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2026 |
| — | Alibaba | Qwen3.8 Max (0902) | 1,664 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2026 |
| — | SpaceXAI | Grok 4.6 (xhigh) | 1,663 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | Meta | Muse Spark 1.3 (xhigh) | 1,662 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2026 |
| — | Z AI | GLM 5.3 Flash | 1,655 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback) | 1,650 历史裁判协议 v4.3.1 | -18 / +18 | Sep 2026 |
| — | Alibaba | Qwen3.8-Flash-Next | 1,647 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2026 |
| — | SpaceXAI | Grok 4.6 (high) | 1,643 历史裁判协议 v4.3.1 | -14 / +14 | Aug 2026 |
| — | SpaceXAI | Grok 4.6 (medium) | 1,643 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | Z AI | GLM-5.3 (max) | 1,634 历史裁判协议 v4.3.1 | -14 / +14 | Aug 2026 |
| — | DeepSeek | DeepSeek V4.1 Flash (Reasoning, Max Effort) | 1,632 历史裁判协议 v4.3.1 | -21 / +21 | Sep 2026 |
| — | Anthropic | Claude Fable 5 | 1,631 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | Alibaba | Qwen3.8 Max | 1,630 历史裁判协议 v4.3.1 | -14 / +14 | Aug 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, High Effort) | 1,629 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | Alibaba | Qwen3.8 2.4T A95B | 1,628 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | OpenAI | GPT-5.6 Sol (max) | 1,587 历史裁判协议 v4.3.1 | -13 / +13 | Jul 2026 |
| — | OpenAI | GPT-5.6 Sol (xhigh) | 1,585 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | OpenAI | GPT-6 Astra (max) | 1,580 历史裁判协议 v4.3.1 | -20 / +20 | Sep 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback) | 1,579 历史裁判协议 v4.3.1 | -18 / +18 | Sep 2026 |
| — | DeepSeek | DeepSeek V4 Flash Vision (Reasoning, Max Effort) | 1,577 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | Sapiens AI | Agnes 3.0 Flash | 1,573 历史裁判协议 v4.3.1 | -22 / +22 | Sep 2026 |
| — | OpenAI | GPT-6 Astra (xhigh) | 1,555 历史裁判协议 v4.3.1 | -20 / +20 | Sep 2026 |
| — | Kimi | Kimi K3 (max) | 1,548 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | OpenAI | GPT-6 Astra (Non-reasoning) 本轮来源未列出，保留上次观测 | 1,543 历史参考 2026-09-12 | -20 / +20 历史参考 2026-09-12 | Sep 2026 |
| — | OpenAI | GPT-6 Astra (high) | 1,530 历史裁判协议 v4.3.1 | -20 / +20 | Sep 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Medium Effort) | 1,525 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | OpenAI | GPT-5.6 Sol (high) | 1,524 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Meta | Muse Spark 1.2 (xhigh) | 1,523 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | Anthropic | Claude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback) | 1,504 历史裁判协议 v4.3.1 | -18 / +18 | Sep 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Max Effort) | 1,501 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | OpenAI | GPT-6 Astra (medium) | 1,501 历史裁判协议 v4.3.1 | -20 / +20 | Sep 2026 |
| — | DeepSeek | DeepSeek V4 Pro 0813 (Reasoning, Max Effort) | 1,493 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | Anthropic | Claude Opus 4.8 (Adaptive Reasoning, Max Effort) | 1,489 历史裁判协议 v4.3.1 | -13 / +13 | May 2026 |
| — | OpenAI | GPT-5.6 Terra (xhigh) | 1,479 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | OpenAI | GPT-5.6 Terra (max) | 1,477 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Gemini 3.8 Flash (high) | 1,464 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2026 | |
| — | Alibaba | Qwen3.8 27B (xhigh) | 1,463 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2026 |
| — | SpaceXAI | Grok 4.6 (low) | 1,461 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 |
| — | OpenAI | GPT-5.6 Luna (max) | 1,456 历史裁判协议 v4.3.1 | -13 / +13 | Jul 2026 |
| — | Gemini 3.8 Flash (medium) | 1,455 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2026 | |
| — | DeepSeek | DeepSeek V4 Flash 0731 (Reasoning, Max Effort) | 1,442 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Gemini 3.7 Flash (high) | 1,433 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2026 | |
| — | Alibaba | Qwen3.8 27B (medium) | 1,430 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2026 |
| — | SpaceXAI | Grok 4.5 (high) | 1,430 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | OpenAI | GPT-5.6 Luna (xhigh) | 1,428 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | OpenAI | GPT-5.6 Sol (medium) | 1,427 历史裁判协议 v4.3.1 | -13 / +13 | Jul 2026 |
| — | OpenAI | GPT-6 Astra (low) | 1,419 历史裁判协议 v4.3.1 | -20 / +20 | Sep 2026 |
| — | OpenAI | GPT-5.6 Terra (high) | 1,415 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Gemini 3.7 Flash (medium) | 1,407 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2026 | |
| — | Z AI | GLM-5.2 (max) | 1,406 历史裁判协议 v4.3.1 | -13 / +13 | Jun 2026 |
| — | Alibaba | Qwen3.8 27B (low) | 1,403 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Xhigh Effort) | 1,401 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | Institute of Foundation Models | K2 Horizon 375B A23B | 1,400 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2026 |
| — | OpenAI | GPT-5.5 (xhigh) | 1,396 历史裁判协议 v4.3.1 | -13 / +13 | Apr 2026 |
| — | Anthropic | Claude Opus 4.7 (Adaptive Reasoning, Max Effort) | 1,396 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Sapiens AI | Agnes 2.5 Pro Beta | 1,376 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2026 |
| — | OpenAI | GPT-5.6 Luna (high) | 1,375 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Anthropic | Claude Opus 5 (Adaptive Reasoning, Low Effort) | 1,372 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | OpenAI | GPT-5.5 (high) | 1,372 历史裁判协议 v4.3.1 | -13 / +13 | Apr 2026 |
| — | Gemini 3.8 Flash (low) | 1,359 历史裁判协议 v4.3.1 | -16 / +16 | Sep 2026 | |
| — | OpenAI | GPT-5.6 Sol (low) | 1,355 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Gemini 3.7 Flash (low) | 1,354 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2026 | |
| — | Gemini 3.6 Flash (high) | 1,331 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 | |
| — | OpenAI | GPT-5.6 Terra (medium) | 1,318 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, High Effort) | 1,308 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | OpenAI | GPT-5.4 (xhigh) | 1,307 历史裁判协议 v4.3.1 | -14 / +14 | Mar 2026 |
| — | Z AI | GLM-5.2 (Non-reasoning) | 1,306 历史裁判协议 v4.3.1 | -16 / +16 | Jun 2026 |
| — | MiniMax | MiniMax-M3 | 1,304 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | Anthropic | Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) | 1,295 历史裁判协议 v4.3.1 | -14 / +14 | Feb 2026 |
| — | Meta | Muse Spark 1.1 (xhigh) | 1,294 历史裁判协议 v4.3.1 | -16 / +16 | Jul 2026 |
| — | OpenAI | GPT-5.5 (medium) | 1,288 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Anthropic | Claude Sonnet 5 (Non-reasoning, High Effort) | 1,283 历史裁判协议 v4.3.1 | -15 / +15 | Jun 2026 |
| — | OpenAI | GPT-5.6 Sol (Non-reasoning) | 1,273 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Apodex | Apodex 1.1 | 1,273 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2026 |
| — | Gemini 3.5 Flash (high) | 1,259 历史裁判协议 v4.3.1 | -14 / +14 | May 2026 | |
| — | Institute of Foundation Models | K2 Horizon MoVA 36B A4B | 1,243 历史裁判协议 v4.3.1 | -21 / +21 | Sep 2026 |
| — | Multiverse Computing | Quasar 438B (max, based on GLM-5.2) | 1,239 历史裁判协议 v4.3.1 | -19 / +19 | Aug 2026 |
| — | InclusionAI | Ling-3.0-flash-VL | 1,225 历史裁判协议 v4.3.1 | -21 / +21 | Sep 2026 |
| — | DeepSeek | DeepSeek V4 Pro 0424 (Reasoning, Max Effort) | 1,223 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Medium Effort) | 1,223 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | DeepSeek | DeepSeek V4 Pro 0424 (Reasoning, High Effort) | 1,219 历史裁判协议 v4.3.1 | -17 / +17 | Apr 2026 |
| — | Motif Technologies | Motif 3 | 1,199 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2026 |
| — | China Mobile | JT-4.1 Flash 236B A21B | 1,193 历史裁判协议 v4.3.1 | -16 / +16 | Jul 2026 |
| — | Thinking Machines | Inkling Small | 1,191 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2026 |
| — | Alibaba | Qwen3.7 Max | 1,190 历史裁判协议 v4.3.1 | -14 / +14 | May 2026 |
| — | AI9Stars | G9v3-39A5B | 1,189 历史裁判协议 v4.3.1 | -18 / +18 | Aug 2026 |
| — | Xiaomi | MiMo-V2.5-Pro | 1,186 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Kimi | Kimi K3 (low) | 1,185 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2026 |
| — | Motif Technologies | Motif 3 (Beta) | 1,183 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2026 |
| — | Z AI | GLM-5.1 (Reasoning) | 1,180 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | OpenAI | GPT-5.6 Terra (low) | 1,178 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | Nex AGI | Nex-N2-Pro (based on Qwen3.5-397B-A17B) | 1,175 历史裁判协议 v4.3.1 | -15 / +15 | Jun 2026 |
| — | Upstage | Solar Pro 4 | 1,173 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2026 |
| — | InclusionAI | Ling-3.0-flash-Fin | 1,171 历史裁判协议 v4.3.1 | -21 / +21 | Sep 2026 |
| — | Thinking Machines | Inkling (xhigh) | 1,165 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2026 |
| — | OpenAI | GPT-5.6 Luna (medium) | 1,164 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | OpenAI | GPT-5.6 Terra (Non-reasoning) | 1,146 历史裁判协议 v4.3.1 | -14 / +14 | Jul 2026 |
| — | Anthropic | Claude Sonnet 5 (Adaptive Reasoning, Low Effort) | 1,145 历史裁判协议 v4.3.1 | -15 / +15 | Jun 2026 |
| — | SpaceXAI | Grok Build 0.1 0616 | 1,138 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | Tencent | Hy3 | 1,136 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2026 |
| — | DeepSeek | DeepSeek V4 Flash 0420 (Reasoning, Max Effort) | 1,116 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Kimi | Kimi K2.7 Code | 1,114 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | Alibaba | Qwen3.8 27B (Non-reasoning) | 1,113 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2026 |
| — | OpenAI | GPT-5.5 (low) | 1,113 历史裁判协议 v4.3.1 | -15 / +15 | Apr 2026 |
| — | Kimi | Kimi K2.6 | 1,096 历史裁判协议 v4.3.1 | -13 / +13 | Apr 2026 |
| — | Z AI | GLM-4.7 (Reasoning) | 1,096 历史裁判协议 v4.3.1 | -16 / +16 | Dec 2025 |
| — | OpenAI | GPT-5.4 mini (xhigh) | 1,095 历史裁判协议 v4.3.1 | -14 / +14 | Mar 2026 |
| — | Institute of Foundation Models | K2 Horizon 7B | 1,091 历史裁判协议 v4.3.1 | -21 / +21 | Sep 2026 |
| — | NVIDIA | Nemotron 3 Ultra 550B A55B (Reasoning) | 1,091 历史裁判协议 v4.3.1 | -14 / +14 | Jun 2026 |
| — | Sapiens AI | Agnes 2.5 Pro Alpha | 1,088 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2026 |
| — | MiniMax | MiniMax-M2.7 | 1,087 历史裁判协议 v4.3.1 | -14 / +14 | Mar 2026 |
| — | DeepSeek | DeepSeek V4 Flash 0420 (Reasoning, High Effort) | 1,083 历史裁判协议 v4.3.1 | -17 / +17 | Apr 2026 |
| — | Xiaomi | MiMo-V2.5 | 1,079 历史裁判协议 v4.3.1 | -17 / +17 | Apr 2026 |
| — | Meta | Muse Spark | 1,076 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | OpenAI | GPT-5.6 Luna (low) | 1,075 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | Alibaba | Qwen3.6 27B (Reasoning) | 1,069 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Alibaba | Qwen3.6 Plus | 1,066 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Gemini 3.5 Flash-Lite | 1,063 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2026 | |
| — | OpenAI | GPT-5.5 (Non-reasoning) | 1,054 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Upstage | Solar Open2 250B | 1,053 历史裁判协议 v4.3.1 | -18 / +18 | Aug 2026 |
| — | SK Telecom | A.X-K2 | 1,044 历史裁判协议 v4.3.1 | -18 / +18 | Aug 2026 |
| — | Alibaba | Qwen3.6 27B (Non-reasoning) | 1,043 历史裁判协议 v4.3.1 | -16 / +16 | Apr 2026 |
| — | OpenAI | GPT-5.4 nano (xhigh) | 1,035 历史裁判协议 v4.3.1 | -14 / +14 | Mar 2026 |
| — | SpaceXAI | Grok 4.3 (Non-reasoning) | 1,031 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | SpaceXAI | Grok 4.3 (high) | 1,018 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | InclusionAI | Ling 3.0 Flash | 1,017 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2026 |
| — | OpenAI | GPT-5 (high) | 1,015 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2025 |
| — | Alibaba | Qwen3.6 35B A3B (Reasoning) | 991 历史裁判协议 v4.3.1 | -14 / +14 | Apr 2026 |
| — | Anthropic | Claude 4.5 Sonnet (Reasoning) | 989 历史裁判协议 v4.3.1 | -16 / +16 | Sep 2025 |
| — | OpenAI | GPT-5.6 Luna (Non-reasoning) | 987 历史裁判协议 v4.3.1 | -15 / +15 | Jul 2026 |
| — | LongCat | LongCat 2.0 | 965 历史裁判协议 v4.3.1 | -16 / +16 | Jun 2026 |
| — | Institute of Foundation Models | K2 Horizon 3.7B | 962 历史裁判协议 v4.3.1 | -21 / +21 | Sep 2026 |
| — | StepFun | Step 3.7 Flash | 954 历史裁判协议 v4.3.1 | -14 / +14 | May 2026 |
| — | Alibaba | Qwen3.6 35B A3B (Non-reasoning) | 953 历史裁判协议 v4.3.1 | -18 / +18 | Apr 2026 |
| — | Kimi | Kimi K2.5 (Reasoning) | 936 历史裁判协议 v4.3.1 | -17 / +17 | Jan 2026 |
| — | OpenAI | GPT-5.1 (high) | 930 历史裁判协议 v4.3.1 | -17 / +17 | Nov 2025 |
| — | Alibaba | Qwen3.5 122B A10B (Reasoning) | 925 历史裁判协议 v4.3.1 | -15 / +15 | Feb 2026 |
| — | LG AI Research | K-EXAONE 2.0 0803 | 918 历史裁判协议 v4.3.1 | -19 / +19 | Aug 2026 |
| — | Alibaba | Qwen3.5 397B A17B (Reasoning) | 905 历史裁判协议 v4.3.1 | -15 / +15 | Feb 2026 |
| — | Gemini 3.1 Pro Preview | 904 历史裁判协议 v4.3.1 | -15 / +15 | Feb 2026 | |
| — | Meta | Muse Glimmer (high) | 893 历史裁判协议 v4.3.1 | -19 / +19 | Aug 2026 |
| — | Alibaba | Qwen3.7 Plus | 886 历史裁判协议 v4.3.1 | -15 / +15 | Jun 2026 |
| — | OpenAI | GPT-5 mini (high) | 880 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2025 |
| — | Mistral | Mistral Medium 3.5 | 875 历史裁判协议 v4.3.1 | -15 / +15 | Apr 2026 |
| — | Z AI | GLM-4.6 (Reasoning) | 874 历史裁判协议 v4.3.1 | -17 / +17 | Sep 2025 |
| — | InclusionAI | Ring-2.6-1T | 864 历史裁判协议 v4.3.1 | -15 / +15 | May 2026 |
| — | Anthropic | Claude 4.5 Haiku (Reasoning) | 854 历史裁判协议 v4.3.1 | -15 / +15 | Oct 2025 |
| — | KwaiKAT | KAT-Coder-Pro V1 | 845 历史裁判协议 v4.3.1 | -16 / +16 | Nov 2025 |
| — | KwaiKAT | KAT Coder Pro V2 | 843 历史裁判协议 v4.3.1 | -18 / +18 | Mar 2026 |
| — | Alibaba | Qwen3.5 122B A10B (Non-reasoning) | 833 历史裁判协议 v4.3.1 | -18 / +18 | Feb 2026 |
| — | OpenBMB | MiniCPM5-2B | 828 历史裁判协议 v4.3.1 | -22 / +22 | Sep 2026 |
| — | DeepSeek | DeepSeek V3.1 Terminus (Reasoning) | 826 历史裁判协议 v4.3.1 | -18 / +18 | Sep 2025 |
| — | Anthropic | Claude 4 Sonnet (Reasoning) | 819 历史裁判协议 v4.3.1 | -17 / +17 | May 2025 |
| — | DeepSeek | DeepSeek V3.2 (Reasoning) | 810 历史裁判协议 v4.3.1 | -19 / +19 | Dec 2025 |
| — | AI9Stars | G9v3-3B | 807 历史裁判协议 v4.3.1 | -26 / +26 | Jul 2026 |
| — | Xiaomi | MiMo-V2-Flash (Non-reasoning) | 788 历史裁判协议 v4.3.1 | -19 / +19 | Dec 2025 |
| — | NVIDIA | Nemotron 3.5 Lightning | 767 历史裁判协议 v4.3.1 | -19 / +19 | Aug 2026 |
| — | Gemma 4 31B (Reasoning) | 755 历史裁判协议 v4.3.1 | -16 / +16 | Apr 2026 | |
| — | OpenAI | gpt-oss-120b (high) | 745 历史裁判协议 v4.3.1 | -15 / +15 | Aug 2025 |
| — | Alibaba | Qwen3.5 35B A3B (Non-reasoning) | 740 历史裁判协议 v4.3.1 | -20 / +20 | Feb 2026 |
| — | OpenAI | GPT-5.4 mini (Non-Reasoning) | 734 历史裁判协议 v4.3.1 | -16 / +16 | Mar 2026 |
| — | IBM | Granite 4.2 30B | 721 历史裁判协议 v4.3.1 | -20 / +20 | Aug 2026 |
| — | InclusionAI | Ling 3.0 Tiny | 716 历史裁判协议 v4.3.1 | -20 / +20 | Aug 2026 |
| — | Gemma 4 26B A4B (Reasoning) | 693 历史裁判协议 v4.3.1 | -15 / +15 | Apr 2026 | |
| — | Gemma 4 31B (Non-reasoning) | 691 历史裁判协议 v4.3.1 | -18 / +18 | Apr 2026 | |
| — | Mistral | Devstral 2 | 689 历史裁判协议 v4.3.1 | -17 / +17 | Dec 2025 |
| — | OpenAI | GPT-5.4 nano (Non-Reasoning) | 681 历史裁判协议 v4.3.1 | -21 / +21 | Mar 2026 |
| — | Mistral | Devstral Small 2 | 680 历史裁判协议 v4.3.1 | -16 / +16 | Dec 2025 |
| — | Alibaba | Qwen3 Coder Next | 664 历史裁判协议 v4.3.1 | -18 / +18 | Feb 2026 |
| — | OpenAI | GPT-5.5 Instant (June 2026) | 662 历史裁判协议 v4.3.1 | -17 / +17 | Jun 2026 |
| — | Cohere | Command A+ | 658 历史裁判协议 v4.3.1 | -18 / +18 | May 2026 |
| — | Inception | Mercury 2 | 646 历史裁判协议 v4.3.1 | -19 / +19 | Feb 2026 |
| — | IBM | Granite 4.2 8B | 645 历史裁判协议 v4.3.1 | -21 / +21 | Aug 2026 |
| — | NVIDIA | Nemotron 3 Super 120B A12B (Reasoning) | 644 历史裁判协议 v4.3.1 | -16 / +16 | Mar 2026 |
| — | Amazon | Nova 2.0 Pro Preview (medium) | 630 历史裁判协议 v4.3.1 | -16 / +16 | Nov 2025 |
| — | LG AI Research | EXAONE 4.5 33B | 628 历史裁判协议 v4.3.1 | -20 / +20 | Apr 2026 |
| — | Gemini 2.5 Pro | 616 历史裁判协议 v4.3.1 | -17 / +17 | Jun 2025 | |
| — | Multiverse Computing | HyperNova 60B 2605 (high, based on gpt-oss-120b) | 604 历史裁判协议 v4.3.1 | -19 / +19 | May 2026 |
| — | Amazon | Nova 2.0 Pro Preview (low) | 601 历史裁判协议 v4.3.1 | -16 / +16 | Nov 2025 |
| — | Gemini 3.1 Flash-Lite | 596 历史裁判协议 v4.3.1 | -16 / +16 | Mar 2026 | |
| — | Gemma 4 12B (Reasoning) | 591 历史裁判协议 v4.3.1 | -20 / +20 | Jun 2026 | |
| — | Alibaba | Qwen3.5 9B (Reasoning) | 588 历史裁判协议 v4.3.1 | -18 / +18 | Mar 2026 |
| — | Mistral | Mistral Large 3 | 588 历史裁判协议 v4.3.1 | -16 / +16 | Dec 2025 |
| — | Mistral | Mistral Small 3.1 | 550 历史裁判协议 v4.3.1 | -18 / +18 | Mar 2025 |
| — | LG AI Research | K-EXAONE (Reasoning) | 540 历史裁判协议 v4.3.1 | -20 / +20 | Dec 2025 |
| — | Mistral | Mistral Small 4 (Reasoning) | 537 历史裁判协议 v4.3.1 | -19 / +19 | Mar 2026 |
| — | Mistral | Mistral Medium 3.1 | 536 历史裁判协议 v4.3.1 | -17 / +17 | Aug 2025 |
| — | Amazon | Nova 2.0 Lite (high) | 536 历史裁判协议 v4.3.1 | -17 / +17 | Oct 2025 |
| — | OpenAI | gpt-oss-20b (high) | 514 历史裁判协议 v4.3.1 | -16 / +16 | Aug 2025 |
| — | Arcee AI | Trinity Large Thinking | 512 历史裁判协议 v4.3.1 | -20 / +20 | Apr 2026 |
| — | Amazon | Nova 2.0 Pro Preview (Non-reasoning) | 509 历史裁判协议 v4.3.1 | -16 / +16 | Nov 2025 |
| — | DiffusionGemma 26B A4B | 501 历史裁判协议 v4.3.1 | -17 / +17 | Jun 2026 | |
| — | Alibaba | Qwen3 235B A22B 2507 (Reasoning) | 495 历史裁判协议 v4.3.1 | -19 / +19 | Jul 2025 |
| — | InclusionAI | Ling 2.6 Flash | 493 历史裁判协议 v4.3.1 | -19 / +19 | Apr 2026 |
| — | Celeris | Celeris-1 | 483 历史裁判协议 v4.3.1 | -21 / +21 | Jul 2026 |
| — | DeepSeek | DeepSeek R1 (Jan '25) | 479 历史裁判协议 v4.3.1 | -18 / +18 | Jan 2025 |
| — | Cohere | North Mini Code | 472 历史裁判协议 v4.3.1 | -18 / +18 | Jun 2026 |
| — | OpenAI | GPT-4.1 mini | 453 历史裁判协议 v4.3.1 | -18 / +18 | Apr 2025 |
| — | NVIDIA | Nemotron Cascade 2 30B A3B | 447 历史裁判协议 v4.3.1 | -21 / +21 | Mar 2026 |
| — | Upstage | Solar Pro 3 | 447 历史裁判协议 v4.3.1 | -16 / +16 | Apr 2026 |
| — | NVIDIA | NVIDIA Nemotron 3 Nano 30B A3B (Reasoning) | 438 历史裁判协议 v4.3.1 | -17 / +17 | Dec 2025 |
| — | Mistral | Ministral 3 14B | 432 历史裁判协议 v4.3.1 | -17 / +17 | Dec 2025 |
| — | OpenAI | o3-mini (high) | 421 历史裁判协议 v4.3.1 | -19 / +19 | Jan 2025 |
| — | NVIDIA | Nemotron 3 Nano Omni 30B A3B Reasoning | 416 历史裁判协议 v4.3.1 | -21 / +21 | Apr 2026 |
| — | Anthropic | Claude 3.5 Haiku | 408 历史裁判协议 v4.3.1 | -18 / +18 | Oct 2024 |
| — | Mistral | Ministral 3 8B | 405 历史裁判协议 v4.3.1 | -17 / +17 | Dec 2025 |
| — | IBM | Granite 4.2 3B | 390 历史裁判协议 v4.3.1 | -21 / +21 | Aug 2026 |
| — | IBM | Granite 4.1 30B | 379 历史裁判协议 v4.3.1 | -16 / +16 | Apr 2026 |
| — | OpenAI | gpt-oss-120b (low) | 364 历史裁判协议 v4.3.1 | -21 / +21 | Aug 2025 |
| — | Mistral | Magistral Medium 1.2 | 362 历史裁判协议 v4.3.1 | -18 / +18 | Sep 2025 |
| — | Institute of Foundation Models | K2 Think V2 | 328 历史裁判协议 v4.3.1 | -18 / +18 | Dec 2025 |
| — | Alibaba | Qwen3 Next 80B A3B (Reasoning) | 325 历史裁判协议 v4.3.1 | -20 / +20 | Sep 2025 |
| — | DeepSeek | DeepSeek V3 0324 | 275 历史裁判协议 v4.3.1 | -19 / +19 | Mar 2025 |
| — | Alibaba | Qwen3 30B A3B 2507 (Reasoning) | 274 历史裁判协议 v4.3.1 | -20 / +20 | Jul 2025 |
| — | Alibaba | Qwen3 32B (Reasoning) | 240 历史裁判协议 v4.3.1 | -18 / +18 | Apr 2025 |
| — | Mistral | Ministral 3 3B | 234 历史裁判协议 v4.3.1 | -18 / +18 | Dec 2025 |
| — | Mistral | Magistral Small 1.2 | 213 历史裁判协议 v4.3.1 | -19 / +19 | Sep 2025 |
| — | Liquid AI | LFM2.5-2.6B | 202 历史裁判协议 v4.3.1 | -22 / +22 | Aug 2026 |
| — | OpenAI | GPT-4 | 189 历史裁判协议 v4.3.1 | -18 / +18 | Mar 2023 |
| — | OpenAI | GPT-4o mini | 188 历史裁判协议 v4.3.1 | -19 / +19 | Jul 2024 |
| — | Alibaba | Qwen3 14B (Reasoning) | 188 历史裁判协议 v4.3.1 | -19 / +19 | Apr 2025 |
| — | DeepSeek | DeepSeek V3 (Dec '24) | 183 历史裁判协议 v4.3.1 | -19 / +19 | Dec 2024 |
| — | Gemma 4 E4B (Reasoning) | 177 历史裁判协议 v4.3.1 | -22 / +22 | Apr 2026 | |
| — | Alibaba | Qwen3.5 2B (Reasoning) | 166 历史裁判协议 v4.3.1 | -21 / +21 | Mar 2026 |
| — | Alibaba | Qwen3 8B (Reasoning) | 165 历史裁判协议 v4.3.1 | -19 / +19 | Apr 2025 |
| — | NVIDIA | NVIDIA Nemotron 3 Nano 4B | 159 历史裁判协议 v4.3.1 | -21 / +21 | Mar 2026 |
| — | IBM | Granite 4.1 3B | 74 历史裁判协议 v4.3.1 | -20 / +20 | Apr 2026 |
| — | Meta | Llama 4 Scout | 60 历史裁判协议 v4.3.1 | -17 / +17 | Apr 2025 |
| — | Meta | Llama 3.3 Instruct 70B | 50 历史裁判协议 v4.3.1 | -19 / +19 | Dec 2024 |
| — | Gemma 4 E2B (Reasoning) | 36 历史裁判协议 v4.3.1 | -21 / +21 | Apr 2026 | |
| — | Mistral | Mistral Small 3.2 | 29 历史裁判协议 v4.3.1 | -19 / +19 | Jun 2025 |
| — | OpenAI | GPT-4.1 nano | 13 历史裁判协议 v4.3.1 | -19 / +19 | Apr 2025 |
| — | Meta | Llama 4 Maverick | -45 历史裁判协议 v4.3.1 | -17 / +17 | Apr 2025 |
| — | NVIDIA | NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning) | -122 历史裁判协议 v4.3.1 | -17 / +17 | Dec 2025 |
| — | Alibaba | Qwen3.5 2B (Non-reasoning) | -130 历史裁判协议 v4.3.1 | -18 / +18 | Mar 2026 |
| — | Alibaba | Qwen3.5 0.8B (Non-reasoning) | -133 历史裁判协议 v4.3.1 | -18 / +18 | Mar 2026 |
| — | OpenBMB | MiniCPM-V 4.6 1.3B | -134 历史裁判协议 v4.3.1 | -17 / +17 | May 2026 |
| — | Meta | Llama 3.1 Instruct 8B | -151 历史裁判协议 v4.3.1 | -17 / +17 | Jul 2024 |
| — | Alibaba | Qwen3.5 0.8B (Reasoning) | -152 历史裁判协议 v4.3.1 | -17 / +17 | Mar 2026 |
| — | Nanbeige | Nanbeige4.1-3B | -166 历史裁判协议 v4.3.1 | -17 / +17 | Feb 2026 |
| — | Gemma 3 12B Instruct | -170 历史裁判协议 v4.3.1 | -17 / +17 | Mar 2025 | |
| — | Gemma 3 27B Instruct | -171 历史裁判协议 v4.3.1 | -16 / +16 | Mar 2025 | |
| — | Microsoft | Phi-4 Mini Instruct | -171 历史裁判协议 v4.3.1 | -17 / +17 | Feb 2024 |
| — | Institute of Foundation Models | K2 Horizon 0.9B | -174 历史裁判协议 v4.3.1 | -17 / +17 | 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 ↗