Skip to content

Benchmark

HELM MMLU: Human Sexuality - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.21

Score distribution

80.6%
Mean
84%
Median
12.5%
Std dev
70.2%
10th percentile
90.8%
90th percentile
26.7%
Min
26.7%93.9%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.21 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 docs93.9%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs93.9%
3DeepSeek V3· Stanford HELM specialized suites data docs92.4%
4Mistral Large2407· Stanford HELM specialized suites data docs92.4%
5text-unicorn· Stanford HELM specialized suites data docs92.4%
6Claude 3 Opus· Stanford HELM specialized suites data docs90.8%
7GPT 4· Stanford HELM specialized suites data docs90.8%
8GPT 4 1106 Preview· Stanford HELM specialized suites data docs90.8%
9GPT 4osetting not named by the source· Stanford HELM specialized suites data docs90.8%
10Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs90.1%
11GPT 4osetting not named by the source· Stanford HELM specialized suites data docs90.1%
12GPT-4 Turbo· Stanford HELM specialized suites data docs90.1%
13Qwen2 72B Instruct· Stanford HELM specialized suites data docs89.3%
14Claude Haiku 3.5· Stanford HELM specialized suites data docs88.5%
15mixtral-8x22b· Stanford HELM specialized suites data docs88.5%
16Nova Prov1· Stanford HELM specialized suites data docs88.5%
17DBRX-instruct· Stanford HELM specialized suites data docs87.8%
18gemini-1.5-pro-002· Stanford HELM specialized suites data docs87.8%
19Meta Llama 3 70b· Stanford HELM specialized suites data docs87.8%
20Qwen1.5-72B· Stanford HELM specialized suites data docs87.8%
21Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs87.8%
22Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs87%
23Yi-34B· Stanford HELM specialized suites data docs87%
24GPT-4o mini· Stanford HELM specialized suites data docs86.3%
25Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs86.3%
26Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs85.5%
27Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs85.5%
28Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs85.5%
29Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs85.5%
30Claude 2.1· Stanford HELM specialized suites data docs84.7%
31gemini-1.5-flash-002· Stanford HELM specialized suites data docs84.7%
32Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs84.7%
33Snowflake Arctic Instruct· Stanford HELM specialized suites data docs84.7%
34DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs84%
35gemma2-27b· Stanford HELM specialized suites data docs84%
36Llama 2 70b· Stanford HELM specialized suites data docs84%
37Nova Litev1· Stanford HELM specialized suites data docs84%
38Jamba 1.5 Large· Stanford HELM specialized suites data docs83.2%
39Claude 3 Haiku· Stanford HELM specialized suites data docs82.4%
40Mistral Small· Stanford HELM specialized suites data docs82.4%
41Nova Microv1· Stanford HELM specialized suites data docs82.4%
42Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs81.7%
43Claude 3 Sonnet· Stanford HELM specialized suites data docs80.9%
44gemma-2-9b· Stanford HELM specialized suites data docs80.9%
45Jamba 1.5 Mini· Stanford HELM specialized suites data docs80.9%
46Claude Instant 1.2· Stanford HELM specialized suites data docs79.4%
47Jamba Instruct· Stanford HELM specialized suites data docs79.4%
48Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs79.4%
49Command R Plus· Stanford HELM specialized suites data docs78.6%
50GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs78.6%

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