Skip to content

Benchmark

HELM MMLU: High School Biology - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.47

Score distribution

84.2%
Mean
89%
Median
13.1%
Std dev
71.6%
10th percentile
94.5%
90th percentile
33.5%
Min
33.5%95.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.47 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
1gemini-1.5-pro-002· Stanford HELM specialized suites data docs95.8%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs95.2%
3DeepSeek V3· Stanford HELM specialized suites data docs94.8%
4GPT 4· Stanford HELM specialized suites data docs94.8%
5GPT 4osetting not named by the source· Stanford HELM specialized suites data docs94.8%
6GPT 4osetting not named by the source· Stanford HELM specialized suites data docs94.5%
7Nova Prov1· Stanford HELM specialized suites data docs94.5%
8Qwen2 72B Instruct· Stanford HELM specialized suites data docs94.5%
9Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs94.2%
10Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs93.9%
11GPT 4 1106 Preview· Stanford HELM specialized suites data docs93.9%
12GPT-4 Turbo· Stanford HELM specialized suites data docs93.9%
13Claude 3 Opus· Stanford HELM specialized suites data docs93.5%
14Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs93.2%
15Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs93.2%
16Mistral Large2407· Stanford HELM specialized suites data docs93.2%
17Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs93.2%
18text-unicorn· Stanford HELM specialized suites data docs93.2%
19gemini-1.5-flash-001· Stanford HELM specialized suites data docs92.9%
20gemini-1.5-pro-001· Stanford HELM specialized suites data docs92.6%
21Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs92.3%
22Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs92.3%
23GPT-4o mini· Stanford HELM specialized suites data docs91.9%
24Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs91.9%
25gemma2-27b· Stanford HELM specialized suites data docs91%
26mixtral-8x22b· Stanford HELM specialized suites data docs91%
27Jamba 1.5 Large· Stanford HELM specialized suites data docs90.6%
28Claude Haiku 3.5· Stanford HELM specialized suites data docs90.3%
29Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs90.3%
30Meta Llama 3 70b· Stanford HELM specialized suites data docs90%
31gemma-2-9b· Stanford HELM specialized suites data docs89.4%
32Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs89.4%
33Yi-34B· Stanford HELM specialized suites data docs89.4%
34Nova Litev1· Stanford HELM specialized suites data docs89%
35Claude 3 Sonnet· Stanford HELM specialized suites data docs88.7%
36Qwen1.5-72B· Stanford HELM specialized suites data docs88.7%
37Claude 2.1· Stanford HELM specialized suites data docs88.4%
38DBRX-instruct· Stanford HELM specialized suites data docs86.8%
39Mistral Small· Stanford HELM specialized suites data docs86.5%
40Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs86.1%
41Claude 3 Haiku· Stanford HELM specialized suites data docs85.5%
42gemini-1.5-flash-002· Stanford HELM specialized suites data docs85.5%
43Qwen1.5-14B· Stanford HELM specialized suites data docs84.8%
44Nova Microv1· Stanford HELM specialized suites data docs84.2%
45DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs83.9%
46GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs83.9%
47Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs83.2%
48Claude Instant 1.2· Stanford HELM specialized suites data docs82.3%
49Snowflake Arctic Instruct· Stanford HELM specialized suites data docs82.3%
50Jamba 1.5 Mini· Stanford HELM specialized suites data docs81.9%

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