Skip to content

Benchmark

HELM MMLU: Miscellaneous - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.31

Score distribution

86.7%
Mean
89.4%
Median
10.3%
Std dev
76.5%
10th percentile
94.9%
90th percentile
29.2%
Min
29.2%96.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.31 for chat models, measured over 64 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs96.4%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs96.2%
3gemini-1.5-pro-002· Stanford HELM specialized suites data docs95.9%
4gemini-1.5-pro-001· Stanford HELM specialized suites data docs95.8%
5GPT 4osetting not named by the source· Stanford HELM specialized suites data docs95.8%
6GPT 4osetting not named by the source· Stanford HELM specialized suites data docs95.4%
7Claude 3 Opus· Stanford HELM specialized suites data docs95.1%
8DeepSeek V3· Stanford HELM specialized suites data docs94.9%
9GPT 4· Stanford HELM specialized suites data docs94.9%
10GPT 4 1106 Preview· Stanford HELM specialized suites data docs94.6%
11GPT-4 Turbo· Stanford HELM specialized suites data docs94.5%
12Qwen2 72B Instruct· Stanford HELM specialized suites data docs94.3%
13Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs93.9%
14Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs93.9%
15Mistral Large2407· Stanford HELM specialized suites data docs93.6%
16Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs93.4%
17Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs93.2%
18Jamba 1.5 Large· Stanford HELM specialized suites data docs93.1%
19Qwen1.5-72B· Stanford HELM specialized suites data docs92.1%
20Meta Llama 3 70b· Stanford HELM specialized suites data docs91.7%
21Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs91.4%
22GPT-4o mini· Stanford HELM specialized suites data docs91.3%
23Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs91.3%
24Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs91.3%
25Nova Prov1· Stanford HELM specialized suites data docs91.2%
26DBRX-instruct· Stanford HELM specialized suites data docs91.1%
27Claude Haiku 3.5· Stanford HELM specialized suites data docs90.5%
28DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs90.4%
29Jamba 1.5 Mini· Stanford HELM specialized suites data docs90.2%
30Yi-34B· Stanford HELM specialized suites data docs90.2%
31gemini-1.5-flash-002· Stanford HELM specialized suites data docs90%
32Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs90%
33mixtral-8x22b· Stanford HELM specialized suites data docs89.9%
34Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs89.4%
35text-unicorn· Stanford HELM specialized suites data docs89.4%
36Claude 3 Haiku· Stanford HELM specialized suites data docs89.3%
37GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs89.3%
38GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs89%
39gemini-1.5-flash-001· Stanford HELM specialized suites data docs88.6%
40gemma2-27b· Stanford HELM specialized suites data docs88.5%
41Claude 2.1· Stanford HELM specialized suites data docs88%
42Snowflake Arctic Instruct· Stanford HELM specialized suites data docs87.5%
43Claude 3 Sonnet· Stanford HELM specialized suites data docs87.2%
44Nova Litev1· Stanford HELM specialized suites data docs87.2%
45Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs87.1%
46Jamba Instruct· Stanford HELM specialized suites data docs86.5%
47Open Mistral Nemo· Stanford HELM specialized suites data docs86.1%
48Llama 2 70b· Stanford HELM specialized suites data docs85.7%
49Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs85.2%
50Command R+· Stanford HELM specialized suites data docs84.8%

Showing 1–50 of 68 results.

Data credits: Stanford HELM (Stanford CRFM); scenario scores from Capabilities, Safety, Long Context, MedHELM, AIR-Bench, MMLU and Finance