Skip to content

Benchmark

HELM MMLU: College Medicine - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
1.97

Score distribution

71.5%
Mean
74%
Median
9.9%
Std dev
61.8%
10th percentile
80.3%
90th percentile
26%
Min
26%86.1%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.97 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
1DeepSeek V3· Stanford HELM specialized suites data docs86.1%
2gemini-1.5-pro-002· Stanford HELM specialized suites data docs84.4%
3GPT 4osetting not named by the source· Stanford HELM specialized suites data docs84.4%
4GPT 4osetting not named by the source· Stanford HELM specialized suites data docs83.8%
5Claude 3 Opus· Stanford HELM specialized suites data docs83.2%
6Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs83.2%
7Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs82.7%
8Mistral Large2407· Stanford HELM specialized suites data docs80.3%
9Nova Prov1· Stanford HELM specialized suites data docs80.3%
10Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs80.3%
11Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs79.8%
12Meta Llama 3 70b· Stanford HELM specialized suites data docs79.8%
13Qwen2 72B Instruct· Stanford HELM specialized suites data docs79.8%
14gemini-1.5-pro-001· Stanford HELM specialized suites data docs79.2%
15GPT 4 1106 Preview· Stanford HELM specialized suites data docs79.2%
16GPT-4 Turbo· Stanford HELM specialized suites data docs79.2%
17gemini-1.5-flash-002· Stanford HELM specialized suites data docs78%
18gemma2-27b· Stanford HELM specialized suites data docs78%
19Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs78%
20Qwen1.5-72B· Stanford HELM specialized suites data docs78%
21Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs76.9%
22mixtral-8x22b· Stanford HELM specialized suites data docs76.9%
23text-unicorn· Stanford HELM specialized suites data docs76.9%
24GPT 4· Stanford HELM specialized suites data docs76.3%
25GPT-4o mini· Stanford HELM specialized suites data docs76.3%
26Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs76.3%
27Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs75.7%
28gemini-1.5-flash-001· Stanford HELM specialized suites data docs75.1%
29Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs75.1%
30Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs74.6%
31Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs74.6%
32Claude 3 Sonnet· Stanford HELM specialized suites data docs74%
33DBRX-instruct· Stanford HELM specialized suites data docs74%
34gemma-7b· Stanford HELM specialized suites data docs74%
35Nova Litev1· Stanford HELM specialized suites data docs74%
36Claude 2.1· Stanford HELM specialized suites data docs73.4%
37Claude 3 Haiku· Stanford HELM specialized suites data docs73.4%
38Claude Haiku 3.5· Stanford HELM specialized suites data docs72.3%
39Jamba 1.5 Large· Stanford HELM specialized suites data docs72.3%
40Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs71.1%
41gemma-2-9b· Stanford HELM specialized suites data docs70.5%
42Yi-34B· Stanford HELM specialized suites data docs70.5%
43DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs69.9%
44Snowflake Arctic Instruct· Stanford HELM specialized suites data docs69.9%
45Qwen1.5-14B· Stanford HELM specialized suites data docs69.4%
46Command R Plus· Stanford HELM specialized suites data docs68.8%
47Claude Instant 1.2· Stanford HELM specialized suites data docs68.2%
48Open Mistral Nemo· Stanford HELM specialized suites data docs68.2%
49Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs67.6%
50Llama-3.2-11B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs67.1%

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