Skip to content

Benchmark

HELM MMLU: Public Relations - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.12

Score distribution

72.4%
Mean
73.6%
Median
7.7%
Std dev
64.5%
10th percentile
80%
90th percentile
34.5%
Min
34.5%85.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.12 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 docs85.5%
2Claude 3 Opus· Stanford HELM specialized suites data docs82.7%
3gemini-1.5-pro-001· Stanford HELM specialized suites data docs81.8%
4Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs81.8%
5DeepSeek V3· Stanford HELM specialized suites data docs80.9%
6gemini-1.5-pro-002· Stanford HELM specialized suites data docs80.9%
7GPT 4osetting not named by the source· Stanford HELM specialized suites data docs80.9%
8Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs80%
9Nova Prov1· Stanford HELM specialized suites data docs80%
10Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs79.1%
11GPT-4o mini· Stanford HELM specialized suites data docs79.1%
12Claude 3 Sonnet· Stanford HELM specialized suites data docs78.2%
13GPT 4 1106 Preview· Stanford HELM specialized suites data docs78.2%
14GPT 4osetting not named by the source· Stanford HELM specialized suites data docs78.2%
15Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs78.2%
16Claude 2.1· Stanford HELM specialized suites data docs77.3%
17Mistral Small· Stanford HELM specialized suites data docs77.3%
18Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs77.3%
19text-unicorn· Stanford HELM specialized suites data docs77.3%
20gemini-1.5-flash-001· Stanford HELM specialized suites data docs76.4%
21Mistral Large2407· Stanford HELM specialized suites data docs76.4%
22Claude 3 Haiku· Stanford HELM specialized suites data docs75.5%
23GPT-4 Turbo· Stanford HELM specialized suites data docs75.5%
24Jamba 1.5 Large· Stanford HELM specialized suites data docs75.5%
25mixtral-8x22b· Stanford HELM specialized suites data docs75.5%
26Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs75.5%
27Qwen1.5-72B· Stanford HELM specialized suites data docs75.5%
28gemma2-27b· Stanford HELM specialized suites data docs74.5%
29GPT 4· Stanford HELM specialized suites data docs74.5%
30GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs74.5%
31Llama 2 70b· Stanford HELM specialized suites data docs74.5%
32Qwen2 72B Instruct· Stanford HELM specialized suites data docs74.5%
33Yi-34B· Stanford HELM specialized suites data docs74.5%
34gemma-2-9b· Stanford HELM specialized suites data docs73.6%
35Meta Llama 3 8b· Stanford HELM specialized suites data docs73.6%
36Claude Haiku 3.5· Stanford HELM specialized suites data docs72.7%
37GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs72.7%
38Jamba 1.5 Mini· Stanford HELM specialized suites data docs72.7%
39Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs72.7%
40Meta Llama 3 70b· Stanford HELM specialized suites data docs72.7%
41Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs72.7%
42Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs71.8%
43Open Mistral Nemo· Stanford HELM specialized suites data docs71.8%
44Yi-6B· Stanford HELM specialized suites data docs71.8%
45Command R Plus· Stanford HELM specialized suites data docs70.9%
46Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs70.9%
47Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs70.9%
48Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs70.9%
49Command R+· Stanford HELM specialized suites data docs70%
50DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs70%

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