Skip to content

Benchmark

HELM MMLU: Professional Law - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
1.89

Score distribution

57%
Mean
59.8%
Median
11.6%
Std dev
42.4%
10th percentile
70.8%
90th percentile
23.8%
Min
23.8%75.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.89 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 docs75.4%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs74.7%
3GPT 4osetting not named by the source· Stanford HELM specialized suites data docs74.2%
4GPT 4· Stanford HELM specialized suites data docs72.7%
5GPT 4osetting not named by the source· Stanford HELM specialized suites data docs72.3%
6Claude 3 Opus· Stanford HELM specialized suites data docs71.5%
7DeepSeek V3· Stanford HELM specialized suites data docs71.3%
8gemini-1.5-pro-002· Stanford HELM specialized suites data docs70.8%
9GPT 4 1106 Preview· Stanford HELM specialized suites data docs70.7%
10Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs70.4%
11Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs67.7%
12Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs67.6%
13Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs67.4%
14GPT-4 Turbo· Stanford HELM specialized suites data docs67.1%
15Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs66.8%
16gemini-1.5-pro-001· Stanford HELM specialized suites data docs66.6%
17Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs66.2%
18Qwen2 72B Instruct· Stanford HELM specialized suites data docs66.2%
19Claude 3 Sonnet· Stanford HELM specialized suites data docs64.5%
20Mistral Large2407· Stanford HELM specialized suites data docs64.5%
21text-unicorn· Stanford HELM specialized suites data docs63.3%
22Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs62.8%
23Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs62.4%
24Meta Llama 3 70b· Stanford HELM specialized suites data docs61.5%
25gemma2-27b· Stanford HELM specialized suites data docs61.1%
26Qwen1.5-72B· Stanford HELM specialized suites data docs60.9%
27Jamba 1.5 Large· Stanford HELM specialized suites data docs60.8%
28Claude 2.1· Stanford HELM specialized suites data docs60.7%
29gemini-1.5-flash-001· Stanford HELM specialized suites data docs60.5%
30mixtral-8x22b· Stanford HELM specialized suites data docs60.5%
31gemini-1.5-flash-002· Stanford HELM specialized suites data docs60.3%
32Nova Prov1· Stanford HELM specialized suites data docs60%
33Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs60%
34Yi-34B· Stanford HELM specialized suites data docs59.8%
35GPT-4o mini· Stanford HELM specialized suites data docs59.2%
36DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs59.1%
37Claude Haiku 3.5· Stanford HELM specialized suites data docs58.4%
38DBRX-instruct· Stanford HELM specialized suites data docs57.9%
39Claude 3 Haiku· Stanford HELM specialized suites data docs57.7%
40Claude Instant 1.2· Stanford HELM specialized suites data docs57%
41Command R Plus· Stanford HELM specialized suites data docs55.7%
42gemma-2-9b· Stanford HELM specialized suites data docs55%
43Nova Litev1· Stanford HELM specialized suites data docs54.6%
44Llama 2 70b· Stanford HELM specialized suites data docs54.1%
45Jamba 1.5 Mini· Stanford HELM specialized suites data docs53.3%
46GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs52.9%
47Command R+· Stanford HELM specialized suites data docs52.6%
48Snowflake Arctic Instruct· Stanford HELM specialized suites data docs51.1%
49Mistral Small· Stanford HELM specialized suites data docs50.9%
50Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs50.4%

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