Skip to content

Benchmark

HELM MMLU: Jurisprudence - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
2.40

Score distribution

82.2%
Mean
85.2%
Median
9.8%
Std dev
73.1%
10th percentile
88.9%
90th percentile
25%
Min
25%90.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.40 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
1GPT 4osetting not named by the source· Stanford HELM specialized suites data docs90.7%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs89.8%
3DeepSeek V3· Stanford HELM specialized suites data docs89.8%
4Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs89.8%
5gemini-1.5-pro-002· Stanford HELM specialized suites data docs89.8%
6GPT 4osetting not named by the source· Stanford HELM specialized suites data docs89.8%
7Yi-34B· Stanford HELM specialized suites data docs89.8%
8Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs88.9%
9gemini-1.5-flash-001· Stanford HELM specialized suites data docs88.9%
10gemini-1.5-pro-001· Stanford HELM specialized suites data docs88.9%
11GPT 4· Stanford HELM specialized suites data docs88.9%
12GPT 4 1106 Preview· Stanford HELM specialized suites data docs88.9%
13Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs88.9%
14Claude 3 Opus· Stanford HELM specialized suites data docs88%
15GPT-4 Turbo· Stanford HELM specialized suites data docs88%
16Jamba 1.5 Mini· Stanford HELM specialized suites data docs88%
17Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs88%
18Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs88%
19text-unicorn· Stanford HELM specialized suites data docs88%
20Claude 2.1· Stanford HELM specialized suites data docs87%
21GPT-4o mini· Stanford HELM specialized suites data docs87%
22Jamba 1.5 Large· Stanford HELM specialized suites data docs87%
23Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs87%
24Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs87%
25Qwen2 72B Instruct· Stanford HELM specialized suites data docs87%
26Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs87%
27Claude 3 Haiku· Stanford HELM specialized suites data docs86.1%
28Claude 3 Sonnet· Stanford HELM specialized suites data docs86.1%
29Claude Haiku 3.5· Stanford HELM specialized suites data docs86.1%
30gemma2-27b· Stanford HELM specialized suites data docs86.1%
31Meta Llama 3 70b· Stanford HELM specialized suites data docs86.1%
32Mistral Large2407· Stanford HELM specialized suites data docs86.1%
33Nova Prov1· Stanford HELM specialized suites data docs86.1%
34DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs85.2%
35gemini-1.5-flash-002· Stanford HELM specialized suites data docs85.2%
36mixtral-8x22b· Stanford HELM specialized suites data docs85.2%
37Nova Litev1· Stanford HELM specialized suites data docs85.2%
38DBRX-instruct· Stanford HELM specialized suites data docs84.3%
39Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs84.3%
40Claude Instant 1.2· Stanford HELM specialized suites data docs83.3%
41gemma-2-9b· Stanford HELM specialized suites data docs83.3%
42Mistral Small· Stanford HELM specialized suites data docs83.3%
43Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs83.3%
44Llama 2 70b· Stanford HELM specialized suites data docs82.4%
45Qwen1.5-72B· Stanford HELM specialized suites data docs82.4%
46Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs81.5%
47Nova Microv1· Stanford HELM specialized suites data docs81.5%
48Command R Plus· Stanford HELM specialized suites data docs80.6%
49GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs80.6%
50GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs80.6%

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