The AI tools index that doesn't waste your time.不浪费你时间的 AI 工具索引。
Knowledge · weight 15%知识 · 权重 15%

MMLU-Pro

⚠ saturation watchlist饱和观察名单 official site官网 ↗

What it tests: A harder re-build of the classic MMLU: 12K questions, 10 answer choices instead of 4, more reasoning-heavy, 14 domains. The original MMLU is saturated (>93%) and retired from our scoring.

考什么:经典 MMLU 的加难重制版:1.2 万题、10 个选项(原来 4 个)、更重推理、14 个学科。原版 MMLU 已饱和(>93%),我们不再计入评分。

Why it counts: Still the standard 'general knowledge' yardstick — but top models cluster within a few points, so we watch its saturation closely.

为什么算数:仍是'通用知识'的标准尺子,但头部模型分差已很小——我们已把它列入饱和观察名单,随时准备降权。

Maintainer维护方TIGER-AI-Lab
License许可Apache 2.0,需署名
Reference point (prior-gen SOTA)参考点(上一代 SOTA)85
◈

Standings实时排名

45 models with sourced scores 个模型有溯源成绩
#Model模型Score成绩Index指数Dated日期Measured by测评方
1Gemini 3.1 Pro PreviewGoogle92.6100—3rd-party第三方source ↗×5
2Claude Fable 5Anthropic91.5992026-08-193rd-party第三方source ↗×2
3Claude Opus 4.8Anthropic89.697—3rd-party第三方source ↗
4Qwen3.7 MaxQwen89.697—3rd-party第三方source ↗×3
5Gemini 3.5 FlashGoogle89.5897—3rd-party第三方source ↗
6Qwen3.7 FlashQwen89.31196—3rd-party第三方source ↗
7Grok 4.5xAI89.296—3rd-party第三方source ↗×2
8GPT-5.6 SolOpenAI89.1962026-083rd-party第三方source ↗×2⚠±6.9
9GLM 5.2Z.ai88.9962026-073rd-party第三方source ↗×3
10Qwen3.8 Max PrimeQwen88.696—3rd-party第三方source ↗
11Qwen3.7 PlusQwen88.5962026-083rd-party第三方source ↗×3
12DeepSeek V4 Flash 0731DeepSeek88.195—3rd-party第三方source ↗
13Kimi K3Moonshot AI8895—3rd-party第三方source ↗
14Kimi K2 ThinkingMoonshot AI87.572⚡high95—3rd-party第三方source ↗×2
15Claude Sonnet 5Anthropic87.5595—3rd-party第三方source ↗×4
16DeepSeek V4 Pro 0423DeepSeek87.594—3rd-party第三方source ↗×5
17Nemotron 3 UltraNVIDIA86.894—3rd-party第三方source ↗×3
18GPT-5.6 TerraOpenAI86.794—3rd-party第三方source ↗
19Grok 4.20xAI86.393—3rd-party第三方source ↗×2⚠±8.7
20DeepSeek V4 Flash 0423DeepSeek86.2932026-083rd-party第三方source ↗×5⚠±18.1
21GPT-5.5OpenAI86.193—3rd-party第三方source ↗×2
22GPT-5.6 LunaOpenAI8693—3rd-party第三方source ↗
23Grok 4.3xAI85.893—3rd-party第三方source ↗
24Gemini 3.5 Flash LiteGoogle85.893—3rd-party第三方source ↗
25DeepSeek V3.2DeepSeek8592—3rd-party第三方source ↗×3
26Kimi K2.7 CodeMoonshot AI84.691—3rd-party第三方source ↗
27MiniMax M3MiniMax84.291—3rd-party第三方source ↗
28GPT-5.6 Sol ProOpenAI84912026-06-013rd-party第三方source ↗
29Gemini 3.1 Flash LiteGoogle83902026-03-033rd-party第三方source ↗×2⚠±5
30MiniMax M2.7MiniMax8289—3rd-party第三方source ↗×2
31Claude Opus 5Anthropic81.9789—3rd-party第三方source ↗×4⚠±10.53
32Mistral Medium 3.5Mistral AI81.388—3rd-party第三方source ↗×2⚠±5.3
33Llama 4 MaverickMeta80.587—3rd-party第三方source ↗×5
34gpt-oss-120bOpenAI7985—3rd-party第三方source ↗×3⚠±11
35Command R+ (08-2024)Cohere75.782—3rd-party第三方source ↗×2⚠±32.5
36Sonar ProPerplexity75.582—3rd-party第三方source ↗
37Llama 4 ScoutMeta74.380—vendor-reported厂商自报source ↗
38Mistral LargeMistral AI73.1179—3rd-party第三方source ↗
39GLM 5.1Z.ai72.779—3rd-party第三方source ↗×2⚠±9.3
40Phi 4Microsoft71.477—3rd-party第三方source ↗×2
41Command ACohere71.277—3rd-party第三方source ↗
42Nova Pro 1.0Amazon69.175—3rd-party第三方source ↗
43Claude Haiku 4.5Anthropic6570—3rd-party第三方source ↗×3⚠±15
44Gemini 3.6 FlashGoogle55.3560—3rd-party第三方source ↗×2⚠±33.95
45Mistral Small 4Mistral AI52.957—3rd-party第三方source ↗

How scores become the Intelligence Score →成绩如何合成智能评分 →