Pass IndexThe State of AISign in

Mmlu — Epoch AI

Mmlu — Epoch AI is run by Epoch AI. It has ranked 217 models, of which the catalogue holds 86, scoring from 0.251 to 0.881 on EM.

measured by Epoch AI · the board itself

PlaceModelMetricScore
1stof 217GPT-4o (2024-11-20)EM0.881
2ndof 217Claude 3.5 SonnetEM0.873
3rdof 217DeepSeek-V3EM0.872
8thof 217Llama 3.3 70B InstructEM0.863
10thof 217Qwen2.5 72B InstructEM0.853
12thof 217Phi-4EM0.848
13thof 217Claude 3 OpusEM0.846
14thof 217Meta-Llama-3.1-405B-InstructEM0.845
15thof 217Llama 3.1 405BEM0.844
16thof 217GPT-4o (2024-08-06)EM0.843
17thof 217GPT-4o (2024-05-13)EM0.842
20thof 217GPT-4 (0613)EM0.824
21stof 217Qwen2-72B-InstructEM0.824
22ndof 217Nova ProEM0.82
24thof 217GPT-4o-mini (2024-07-18)EM0.818
25thof 217GPT-4 TurboEM0.813
26thof 217Llama-3.2-90B-Vision-InstructEM0.803
27thof 217Meta-Llama-3.1-70B-InstructEM0.801
28thof 217Mistral Large 2407EM0.8
29thof 217AgentRLEM0.799
30thof 217Gemini 2.0 FlashEM0.797
32ndof 217Meta-Llama-3-70B-InstructEM0.793
33rdof 217Yi LargeEM0.793
35thof 217Qwen 2.5 Coder 32BEM0.791
36thof 217Claude 2.0EM0.785
37thof 217DeepSeek-V2EM0.784
38thof 217Phi-3-medium-128k-instructEM0.78
42ndof 217Mixtral-8x22B-v0.1EM0.778
44thof 217Nova LiteEM0.77
47thof 217Yi-34BEM0.763
49thof 217Claude 3 SonnetEM0.759
50thof 217Gemma 2 27BEM0.757
51stof 217Phi-3-small-8k-instructEM0.757
54thof 217Claude Haiku 3.5EM0.743
56thof 217Claude 3 HaikuEM0.738
57thof 217Claude 2.1EM0.735
58thof 217Yi-34B-ChatEM0.735
62ndof 217Qwen2.5 7B InstructEM0.729
64thof 217gemma-2-9b-itEM0.721
65thof 217GPT-3.5 Turbo (1106)EM0.714
66thof 217Nova MicroEM0.708
67thof 217falcon-180BEM0.706
72ndof 217Llama-2-70b-hfEM0.699
75thof 217c4ai-command-r-plus-08-2024EM0.694
77thof 217GPT-3.5 Turbo (older v0613)EM0.689
80thof 217Llama 3 8B InstructEM0.688
81stof 217Mistral Large (24.02)EM0.688
82ndof 217Phi-3-mini-4k-instructEM0.688
83rdof 217Mistral Small (24.02)EM0.687
85thof 217StableBeluga2EM0.686
86thof 217Yi-9BEM0.684
87thof 217Qwen2.5-Coder-7BEM0.68
90thof 217GPT-3.5 TurboEM0.673
92ndof 217Qwen-14BEM0.663
93rdof 217gemma-7bEM0.661
94thof 217c4ai-command-r-08-2024EM0.652
95thof 217Qwen-14B-ChatEM0.65
98thof 217starcoder2-15bEM0.641
100thof 217Yi-6BEM0.64
109thof 217Mistral-7B-Instruct-v0.2EM0.625
110thof 217Mistral-7B-v0.1EM0.625
118thof 217Llama-2-70b-chatEM0.599
119thof 217Mistral-7B-Instruct-v0.3EM0.599
125thof 217falcon-11BEM0.584
126thof 217phi-2EM0.584
131stof 217falcon-40bEM0.569
134thof 217Llama 3.2 11B Vision InstructEM0.565
136thof 217Llama 3.1 8B InstructEM0.561
138thof 217Llama-2-13bEM0.556
142ndof 217Baichuan2-13B-ChatEM0.551
151stof 217Baichuan-13B-BaseEM0.516
153rdof 217Llama-2-13b-chatEM0.509
158thof 217llama-13bEM0.477
164thof 217Llama-2-7bEM0.458
169thof 217Mistral 7BEM0.45
173rdof 217Baichuan-7BEM0.423
174thof 217gemma-2bEM0.423
175thof 217Qwen2.5-Coder-0.5BEM0.42
178thof 217starcoder2-7bEM0.388
179thof 217phi-1_5EM0.376
180thof 217starcoder2-3bEM0.366
182ndof 217xgen-7b-8k-baseEM0.363
183rdof 217llama-7bEM0.356
189thof 217falcon-7bEM0.35
211thof 217gpt-j-6bEM0.257
214thof 217opt-1.3bEM0.251

Read from the board on 2026-09-16