Skip to content

Benchmark

HELM MMLU: Management - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
1.94

Score distribution

84.3%
Mean
86.4%
Median
9.8%
Std dev
74.8%
10th percentile
92.2%
90th percentile
27.2%
Min
27.2%94.2%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.94 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 Opus· Stanford HELM specialized suites data docs94.2%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs94.2%
3GPT 4osetting not named by the source· Stanford HELM specialized suites data docs94.2%
4Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs93.2%
5GPT 4· Stanford HELM specialized suites data docs93.2%
6Claude 3 Sonnet· Stanford HELM specialized suites data docs92.2%
7gemini-1.5-pro-001· Stanford HELM specialized suites data docs92.2%
8Nova Prov1· Stanford HELM specialized suites data docs92.2%
9GPT 4 1106 Preview· Stanford HELM specialized suites data docs91.3%
10GPT 4osetting not named by the source· Stanford HELM specialized suites data docs91.3%
11Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs91.3%
12Meta Llama 3 70b· Stanford HELM specialized suites data docs91.3%
13Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs91.3%
14Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs91.3%
15DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs90.3%
16DeepSeek V3· Stanford HELM specialized suites data docs90.3%
17gemini-1.5-pro-002· Stanford HELM specialized suites data docs90.3%
18Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs90.3%
19Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs90.3%
20Qwen2 72B Instruct· Stanford HELM specialized suites data docs90.3%
21text-unicorn· Stanford HELM specialized suites data docs90.3%
22gemini-1.5-flash-002· Stanford HELM specialized suites data docs89.3%
23Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs89.3%
24Yi-34B· Stanford HELM specialized suites data docs89.3%
25GPT-4 Turbo· Stanford HELM specialized suites data docs88.3%
26Mistral Large2407· Stanford HELM specialized suites data docs88.3%
27mixtral-8x22b· Stanford HELM specialized suites data docs88.3%
28Claude 3 Haiku· Stanford HELM specialized suites data docs87.4%
29gemma-2-9b· Stanford HELM specialized suites data docs87.4%
30Meta Llama 3 8b· Stanford HELM specialized suites data docs87.4%
31gemini-1.5-flash-001· Stanford HELM specialized suites data docs86.4%
32gemma-7b· Stanford HELM specialized suites data docs86.4%
33gemma2-27b· Stanford HELM specialized suites data docs86.4%
34Jamba 1.5 Large· Stanford HELM specialized suites data docs86.4%
35Nova Litev1· Stanford HELM specialized suites data docs86.4%
36Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs86.4%
37DBRX-instruct· Stanford HELM specialized suites data docs85.4%
38Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs85.4%
39Qwen1.5-72B· Stanford HELM specialized suites data docs85.4%
40Claude Haiku 3.5· Stanford HELM specialized suites data docs84.5%
41GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs84.5%
42GPT-4o mini· Stanford HELM specialized suites data docs84.5%
43Llama 2 70b· Stanford HELM specialized suites data docs84.5%
44Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs84.5%
45Claude Instant 1.2· Stanford HELM specialized suites data docs83.5%
46Command R Plus· Stanford HELM specialized suites data docs83.5%
47GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs83.5%
48Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs83.5%
49Claude 2.1· Stanford HELM specialized suites data docs82.5%
50Jamba 1.5 Mini· Stanford HELM specialized suites data docs82.5%

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