Skip to content

Benchmark

HELM MMLU: High School European History - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.60

Score distribution

80.7%
Mean
84.8%
Median
11.8%
Std dev
71.5%
10th percentile
88.5%
90th percentile
23.6%
Min
23.6%91.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.60 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
1GPT 4osetting not named by the source· Stanford HELM specialized suites data docs91.5%
2GPT 4osetting not named by the source· Stanford HELM specialized suites data docs90.3%
3Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs89.7%
4Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs89.7%
5GPT 4 1106 Preview· Stanford HELM specialized suites data docs89.7%
6Claude 3 Opus· Stanford HELM specialized suites data docs88.5%
7gemini-1.5-pro-002· Stanford HELM specialized suites data docs88.5%
8GPT-4o mini· Stanford HELM specialized suites data docs88.5%
9Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs88.5%
10Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs88.5%
11Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs88.5%
12Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs88.5%
13DeepSeek V3· Stanford HELM specialized suites data docs87.9%
14GPT 4· Stanford HELM specialized suites data docs87.9%
15Meta Llama 3 70b· Stanford HELM specialized suites data docs87.3%
16Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs87.3%
17Nova Prov1· Stanford HELM specialized suites data docs87.3%
18Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs87.3%
19Qwen2 72B Instruct· Stanford HELM specialized suites data docs87.3%
20gemini-1.5-pro-001· Stanford HELM specialized suites data docs86.7%
21GPT-4 Turbo· Stanford HELM specialized suites data docs86.7%
22Yi-34B· Stanford HELM specialized suites data docs86.7%
23Claude Haiku 3.5· Stanford HELM specialized suites data docs86.1%
24Jamba 1.5 Large· Stanford HELM specialized suites data docs86.1%
25Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs86.1%
26Mistral Large2407· Stanford HELM specialized suites data docs86.1%
27Qwen1.5-72B· Stanford HELM specialized suites data docs86.1%
28text-unicorn· Stanford HELM specialized suites data docs86.1%
29Nova Litev1· Stanford HELM specialized suites data docs85.5%
30Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs85.5%
31Claude 2.1· Stanford HELM specialized suites data docs84.8%
32Claude 3 Sonnet· Stanford HELM specialized suites data docs84.8%
33Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs84.8%
34mixtral-8x22b· Stanford HELM specialized suites data docs84.8%
35Qwen1.5-14B· Stanford HELM specialized suites data docs84.8%
36DBRX-instruct· Stanford HELM specialized suites data docs84.2%
37Jamba 1.5 Mini· Stanford HELM specialized suites data docs83.6%
38Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs83.6%
39Command R+· Stanford HELM specialized suites data docs83%
40DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs83%
41gemini-1.5-flash-002· Stanford HELM specialized suites data docs83%
42Command R Plus· Stanford HELM specialized suites data docs82.4%
43gemini-1.5-flash-001· Stanford HELM specialized suites data docs82.4%
44gemma2-27b· Stanford HELM specialized suites data docs82.4%
45Nova Microv1· Stanford HELM specialized suites data docs82.4%
46Claude 3 Haiku· Stanford HELM specialized suites data docs81.8%
47Claude Instant 1.2· Stanford HELM specialized suites data docs81.8%
48Mistral Small· Stanford HELM specialized suites data docs81.8%
49Jamba Instruct· Stanford HELM specialized suites data docs81.2%
50Llama 2 70b· Stanford HELM specialized suites data docs81.2%

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