Skip to content

Benchmark

HELM MMLU: Medical Genetics - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
1.95

Score distribution

81.7%
Mean
85%
Median
12%
Std dev
69%
10th percentile
93%
90th percentile
28%
Min
28%98%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.95 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 docs98%
2GPT 4osetting not named by the source· Stanford HELM specialized suites data docs98%
3Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs96%
4DeepSeek V3· Stanford HELM specialized suites data docs96%
5GPT 4osetting not named by the source· Stanford HELM specialized suites data docs96%
6GPT 4· Stanford HELM specialized suites data docs94%
7Claude 3 Opus· Stanford HELM specialized suites data docs93%
8GPT 4 1106 Preview· Stanford HELM specialized suites data docs93%
9Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs93%
10Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs93%
11gemini-1.5-pro-002· Stanford HELM specialized suites data docs92%
12GPT-4 Turbo· Stanford HELM specialized suites data docs92%
13Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs92%
14Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs92%
15gemini-1.5-pro-001· Stanford HELM specialized suites data docs91%
16Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs91%
17Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs90%
18Mistral Large2407· Stanford HELM specialized suites data docs90%
19Nova Litev1· Stanford HELM specialized suites data docs90%
20Qwen2 72B Instruct· Stanford HELM specialized suites data docs90%
21Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs89%
22gemini-1.5-flash-002· Stanford HELM specialized suites data docs89%
23GPT-4o mini· Stanford HELM specialized suites data docs89%
24Jamba 1.5 Large· Stanford HELM specialized suites data docs89%
25Meta Llama 3 70b· Stanford HELM specialized suites data docs89%
26gemma2-27b· Stanford HELM specialized suites data docs87%
27Nova Prov1· Stanford HELM specialized suites data docs87%
28Qwen1.5-72B· Stanford HELM specialized suites data docs87%
29Yi-34B· Stanford HELM specialized suites data docs87%
30gemini-1.5-flash-001· Stanford HELM specialized suites data docs86%
31DBRX-instruct· Stanford HELM specialized suites data docs85%
32mixtral-8x22b· Stanford HELM specialized suites data docs85%
33Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs85%
34Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs85%
35gemma-2-9b· Stanford HELM specialized suites data docs84%
36Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs84%
37Claude Haiku 3.5· Stanford HELM specialized suites data docs83%
38Meta Llama 3 8b· Stanford HELM specialized suites data docs83%
39text-unicorn· Stanford HELM specialized suites data docs83%
40Nova Microv1· Stanford HELM specialized suites data docs82%
41Claude 2.1· Stanford HELM specialized suites data docs81%
42Command R+· Stanford HELM specialized suites data docs81%
43Claude 3 Haiku· Stanford HELM specialized suites data docs80%
44GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs80%
45Claude 3 Sonnet· Stanford HELM specialized suites data docs79%
46Open Mistral Nemo· Stanford HELM specialized suites data docs78%
47Command R Plus· Stanford HELM specialized suites data docs77%
48Yi-6B· Stanford HELM specialized suites data docs77%
49Qwen1.5-14B· Stanford HELM specialized suites data docs76%
50Snowflake Arctic Instruct· Stanford HELM specialized suites data docs76%

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