Skip to content

Benchmark

HELM MMLU: High School US History - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.38

Score distribution

86%
Mean
90.7%
Median
12.8%
Std dev
78.9%
10th percentile
94.6%
90th percentile
25%
Min
25%96.1%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.38 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 docs96.1%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs95.6%
3GPT 4osetting not named by the source· Stanford HELM specialized suites data docs95.6%
4GPT 4osetting not named by the source· Stanford HELM specialized suites data docs95.1%
5GPT-4 Turbo· Stanford HELM specialized suites data docs95.1%
6Meta Llama 3 70b· Stanford HELM specialized suites data docs95.1%
7GPT 4· Stanford HELM specialized suites data docs94.6%
8Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs94.6%
9Claude 3 Opus· Stanford HELM specialized suites data docs94.1%
10Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs94.1%
11Qwen2 72B Instruct· Stanford HELM specialized suites data docs94.1%
12Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs93.6%
13GPT 4 1106 Preview· Stanford HELM specialized suites data docs93.6%
14text-unicorn· Stanford HELM specialized suites data docs93.6%
15Claude 3 Sonnet· Stanford HELM specialized suites data docs93.1%
16Jamba 1.5 Large· Stanford HELM specialized suites data docs93.1%
17Llama 2 70b· Stanford HELM specialized suites data docs93.1%
18gemini-1.5-pro-002· Stanford HELM specialized suites data docs92.6%
19Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs92.6%
20Qwen1.5-72B· Stanford HELM specialized suites data docs92.6%
21Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs92.6%
22gemini-1.5-pro-001· Stanford HELM specialized suites data docs92.2%
23Mistral Large2407· Stanford HELM specialized suites data docs92.2%
24Nova Prov1· Stanford HELM specialized suites data docs92.2%
25Claude Haiku 3.5· Stanford HELM specialized suites data docs91.7%
26gemma2-27b· Stanford HELM specialized suites data docs91.7%
27mixtral-8x22b· Stanford HELM specialized suites data docs91.2%
28Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs91.2%
29Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs91.2%
30DBRX-instruct· Stanford HELM specialized suites data docs90.7%
31DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs90.7%
32gemini-1.5-flash-001· Stanford HELM specialized suites data docs90.7%
33GPT-4o mini· Stanford HELM specialized suites data docs90.7%
34Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs90.7%
35Nova Litev1· Stanford HELM specialized suites data docs90.7%
36Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs90.7%
37Yi-34B· Stanford HELM specialized suites data docs90.7%
38gemini-1.5-flash-002· Stanford HELM specialized suites data docs89.7%
39Claude 3 Haiku· Stanford HELM specialized suites data docs89.2%
40gemma-2-9b· Stanford HELM specialized suites data docs89.2%
41Jamba 1.5 Mini· Stanford HELM specialized suites data docs88.7%
42Nova Microv1· Stanford HELM specialized suites data docs88.2%
43Claude Instant 1.2· Stanford HELM specialized suites data docs87.3%
44Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs86.8%
45Open Mistral Nemo· Stanford HELM specialized suites data docs85.8%
46Command R+· Stanford HELM specialized suites data docs85.3%
47Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs85.3%
48Jamba Instruct· Stanford HELM specialized suites data docs85.3%
49Claude 2.1· Stanford HELM specialized suites data docs84.8%
50Command R Plus· Stanford HELM specialized suites data docs84.8%

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