Skip to content

Benchmark

HELM MMLU: Philosophy - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.09

Score distribution

78.4%
Mean
80.4%
Median
9.6%
Std dev
67.2%
10th percentile
88.7%
90th percentile
32.5%
Min
32.5%90%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.09 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 docs90%
2DeepSeek V3· Stanford HELM specialized suites data docs90%
3GPT 4osetting not named by the source· Stanford HELM specialized suites data docs90%
4GPT 4osetting not named by the source· Stanford HELM specialized suites data docs89.4%
5Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs89.1%
6Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs89.1%
7Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs88.7%
8gemini-1.5-pro-002· Stanford HELM specialized suites data docs88.7%
9Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs87.8%
10gemini-1.5-pro-001· Stanford HELM specialized suites data docs87.1%
11GPT-4 Turbo· Stanford HELM specialized suites data docs86.8%
12Meta Llama 3 70b· Stanford HELM specialized suites data docs86.5%
13GPT 4· Stanford HELM specialized suites data docs85.9%
14Qwen2 72B Instruct· Stanford HELM specialized suites data docs85.9%
15Claude 3 Sonnet· Stanford HELM specialized suites data docs85.2%
16GPT 4 1106 Preview· Stanford HELM specialized suites data docs85.2%
17gemma2-27b· Stanford HELM specialized suites data docs84.9%
18Jamba 1.5 Large· Stanford HELM specialized suites data docs84.9%
19mixtral-8x22b· Stanford HELM specialized suites data docs84.2%
20Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs83.9%
21Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs83.9%
22text-unicorn· Stanford HELM specialized suites data docs83.6%
23Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs83.3%
24Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs83%
25Qwen1.5-72B· Stanford HELM specialized suites data docs83%
26Mistral Large2407· Stanford HELM specialized suites data docs82.6%
27Nova Prov1· Stanford HELM specialized suites data docs82.6%
28Claude Haiku 3.5· Stanford HELM specialized suites data docs82.3%
29Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs82.3%
30Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs82%
31Yi-34B· Stanford HELM specialized suites data docs82%
32Nova Litev1· Stanford HELM specialized suites data docs81.7%
33Claude 3 Haiku· Stanford HELM specialized suites data docs81.4%
34DBRX-instruct· Stanford HELM specialized suites data docs80.4%
35Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs80.4%
36DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs80.1%
37gemini-1.5-flash-002· Stanford HELM specialized suites data docs79.7%
38Claude 2.1· Stanford HELM specialized suites data docs79.4%
39Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs79.4%
40gemini-1.5-flash-001· Stanford HELM specialized suites data docs79.1%
41Llama 2 70b· Stanford HELM specialized suites data docs79.1%
42gemma-2-9b· Stanford HELM specialized suites data docs77.2%
43GPT-4o mini· Stanford HELM specialized suites data docs77.2%
44Mistral Small· Stanford HELM specialized suites data docs76.5%
45GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs75.9%
46Claude Instant 1.2· Stanford HELM specialized suites data docs75.6%
47Jamba 1.5 Mini· Stanford HELM specialized suites data docs75.2%
48Snowflake Arctic Instruct· Stanford HELM specialized suites data docs75.2%
49Jamba Instruct· Stanford HELM specialized suites data docs74.9%
50GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs74.6%

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