Skip to content

Benchmark

HELM MMLU: College Biology - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.09

Score distribution

83.8%
Mean
88.2%
Median
13%
Std dev
68.1%
10th percentile
94.4%
90th percentile
23.6%
Min
23.6%97.2%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.09 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 Opus· Stanford HELM specialized suites data docs97.2%
2DeepSeek V3· Stanford HELM specialized suites data docs97.2%
3Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs96.5%
4Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs96.5%
5gemini-1.5-pro-002· Stanford HELM specialized suites data docs96.5%
6Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs96.5%
7text-unicorn· Stanford HELM specialized suites data docs95.1%
8GPT 4osetting not named by the source· Stanford HELM specialized suites data docs94.4%
9Meta Llama 3 70b· Stanford HELM specialized suites data docs94.4%
10Nova Prov1· Stanford HELM specialized suites data docs94.4%
11Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs94.4%
12GPT 4osetting not named by the source· Stanford HELM specialized suites data docs93.8%
13Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs93.8%
14Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs93.8%
15Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs93.8%
16GPT 4 1106 Preview· Stanford HELM specialized suites data docs93.1%
17GPT-4 Turbo· Stanford HELM specialized suites data docs93.1%
18Qwen2 72B Instruct· Stanford HELM specialized suites data docs93.1%
19gemini-1.5-flash-001· Stanford HELM specialized suites data docs92.4%
20GPT 4· Stanford HELM specialized suites data docs92.4%
21gemma2-27b· Stanford HELM specialized suites data docs91.7%
22Claude 3 Sonnet· Stanford HELM specialized suites data docs91%
23Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs91%
24GPT-4o mini· Stanford HELM specialized suites data docs91%
25Jamba 1.5 Large· Stanford HELM specialized suites data docs91%
26Nova Litev1· Stanford HELM specialized suites data docs91%
27Qwen1.5-72B· Stanford HELM specialized suites data docs91%
28gemini-1.5-pro-001· Stanford HELM specialized suites data docs90.3%
29mixtral-8x22b· Stanford HELM specialized suites data docs90.3%
30DBRX-instruct· Stanford HELM specialized suites data docs89.6%
31Mistral Large2407· Stanford HELM specialized suites data docs89.6%
32Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs88.9%
33Yi-34B· Stanford HELM specialized suites data docs88.9%
34Claude 2.1· Stanford HELM specialized suites data docs88.2%
35Claude 3 Haiku· Stanford HELM specialized suites data docs88.2%
36gemma-2-9b· Stanford HELM specialized suites data docs88.2%
37Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs88.2%
38DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs85.4%
39Claude Haiku 3.5· Stanford HELM specialized suites data docs84.7%
40Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs84.7%
41Llama 2 70b· Stanford HELM specialized suites data docs84%
42Command R Plus· Stanford HELM specialized suites data docs83.3%
43Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs82.6%
44Nova Microv1· Stanford HELM specialized suites data docs81.3%
45Jamba 1.5 Mini· Stanford HELM specialized suites data docs80.6%
46Mistral Small· Stanford HELM specialized suites data docs80.6%
47Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs79.9%
48gemma-7b· Stanford HELM specialized suites data docs79.2%
49GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs79.2%
50GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs78.5%

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