Skip to content

Benchmark

HELM MMLU: High School Macroeconomics - EM

Scores are percentages. Higher is better.

Models measured
64
Scores recorded
68
Sources
1
IQ weight
1.94

Score distribution

76.4%
Mean
78.7%
Median
11.9%
Std dev
63.3%
10th percentile
89.2%
90th percentile
33.6%
Min
33.6%91.3%
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
1gemini-1.5-pro-002· Stanford HELM specialized suites data docs91.3%
2GPT 4osetting not named by the source· Stanford HELM specialized suites data docs91%
3DeepSeek V3· Stanford HELM specialized suites data docs90.8%
4Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs90.3%
5GPT 4osetting not named by the source· Stanford HELM specialized suites data docs90.3%
6Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs90%
7Gemini 2.0 Flash Experimental· Stanford HELM specialized suites data docs89.7%
8Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs89.2%
9gemini-1.5-pro-001· Stanford HELM specialized suites data docs89%
10Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs88.7%
11Qwen2 72B Instruct· Stanford HELM specialized suites data docs87.9%
12Claude 3 Opus· Stanford HELM specialized suites data docs87.2%
13GPT 4· Stanford HELM specialized suites data docs87.2%
14Nova Prov1· Stanford HELM specialized suites data docs87.2%
15GPT 4 1106 Preview· Stanford HELM specialized suites data docs86.7%
16GPT-4 Turbo· Stanford HELM specialized suites data docs86.2%
17Llama 3.3 70B Instruct Turbo· Stanford HELM specialized suites data docs85.9%
18gemini-1.5-flash-001· Stanford HELM specialized suites data docs85.1%
19GPT-4o mini· Stanford HELM specialized suites data docs85.1%
20Llama-3.2-90B-Vision-Instruct-Turbo· Stanford HELM specialized suites data docs84.1%
21Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs84.1%
22text-unicorn· Stanford HELM specialized suites data docs84.1%
23gemini-1.5-flash-002· Stanford HELM specialized suites data docs83.8%
24Jamba 1.5 Large· Stanford HELM specialized suites data docs83.3%
25Meta Llama 3 70b· Stanford HELM specialized suites data docs83.3%
26Mistral Large2407· Stanford HELM specialized suites data docs83.1%
27Phi-3 Medium 4K Instruct· Stanford HELM specialized suites data docs83.1%
28Nova Litev1· Stanford HELM specialized suites data docs81.3%
29Qwen1.5-72B· Stanford HELM specialized suites data docs81.3%
30Phi-3 Small 8K Instruct· Stanford HELM specialized suites data docs81%
31gemma2-27b· Stanford HELM specialized suites data docs80.8%
32mixtral-8x22b· Stanford HELM specialized suites data docs80.3%
33Yi-34B· Stanford HELM specialized suites data docs80.3%
34Claude 3 Sonnet· Stanford HELM specialized suites data docs78.7%
35DBRX-instruct· Stanford HELM specialized suites data docs78.5%
36Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs78.2%
37gemma-2-9b· Stanford HELM specialized suites data docs77.4%
38Claude 2.1· Stanford HELM specialized suites data docs77.2%
39Qwen1.5-110B-Chat· Stanford HELM specialized suites data docs76.7%
40Claude Haiku 3.5· Stanford HELM specialized suites data docs75.9%
41Jamba 1.5 Mini· Stanford HELM specialized suites data docs75.6%
42DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs74.6%
43Nova Microv1· Stanford HELM specialized suites data docs74.6%
44Mistral Small· Stanford HELM specialized suites data docs73.8%
45Llama 2 70b· Stanford HELM specialized suites data docs73.1%
46Mistral Largesetting not named by the source· Stanford HELM specialized suites data docs73.1%
47Claude 3 Haiku· Stanford HELM specialized suites data docs72.8%
48Qwen1.5-14B· Stanford HELM specialized suites data docs72.8%
49Jamba Instruct· Stanford HELM specialized suites data docs70.3%
50Claude Instant 1.2· Stanford HELM specialized suites data docs69.5%

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