Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - #12: Military and Warfare

Safety benchmarks examine risks in model behaviour, including how models respond to unsafe requests. Scores are percentages. Higher is better.

Models measured
70
Scores recorded
75
Sources
1
IQ weight
2.08

Score distribution

64.5%
Mean
78.3%
Median
31.8%
Std dev
10%
10th percentile
98.3%
90th percentile
8.3%
Min
8.3%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.08 for chat models, measured over 70 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 Haiku 4.5· Stanford HELM specialized suites data docs100%
2GPT 5· Stanford HELM specialized suites data docs100%
3GPT OSS 20b· Stanford HELM specialized suites data docs100%
4GPT-5 mini· Stanford HELM specialized suites data docs100%
5GPT-5 nano· Stanford HELM specialized suites data docs100%
6o1· Stanford HELM specialized suites data docs100%
7Qwen3 Next 80B A3Bthinking· Stanford HELM specialized suites data docs100%
8GPT 5.1· Stanford HELM specialized suites data docs98.3%
9Claude 3 Sonnet· Stanford HELM specialized suites data docs96.7%
10Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs95%
11o4 Mini· Stanford HELM specialized suites data docs95%
12OpenAI o3· Stanford HELM specialized suites data docs93.3%
13Kimi K2 Instruct· Stanford HELM specialized suites data docs91.7%
14Qwen3 235B A22B Instruct 2507· Stanford HELM specialized suites data docs91.7%
15Claude 3 Haiku· Stanford HELM specialized suites data docs90%
16Claude 3 Opus· Stanford HELM specialized suites data docs90%
17GPT OSS 120B· Stanford HELM specialized suites data docs90%
18Gemini 2.5 Pro Preview 03-25· Stanford HELM specialized suites data docs88.3%
19Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs86.7%
20Claude Sonnet 3.7· Stanford HELM specialized suites data docs86.7%
21Claude Sonnet 4.5· Stanford HELM specialized suites data docs86.7%
22Granite 3.3 8b Instruct· Stanford HELM specialized suites data docs86.7%
23Llama3 8B Chat· Stanford HELM specialized suites data docs86.7%
24Palmyra X5· Stanford HELM specialized suites data docs86.7%
25gemini-2.5-flash-preview-04-17· Stanford HELM specialized suites data docs85%
26GPT-4 Turbo· Stanford HELM specialized suites data docs85%
27o3 Mini· Stanford HELM specialized suites data docs85%
28Claude Opus 4· Stanford HELM specialized suites data docs83.3%
29Claude Sonnet 4· Stanford HELM specialized suites data docs83.3%
30GPT 4.1· Stanford HELM specialized suites data docs83.3%
31Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs83.3%
32Llama 4 Maverick· Stanford HELM specialized suites data docs83.3%
33Gemini 3 Pro Preview· Stanford HELM specialized suites data docs81.7%
34GPT 4· Stanford HELM specialized suites data docs81.7%
35GPT-4.5 Preview· Stanford HELM specialized suites data docs81.7%
36gemini-2.0-pro-exp-02-05· Stanford HELM specialized suites data docs80%
37GPT-4o mini· Stanford HELM specialized suites data docs80%
38Gemini 2.0 Flash· Stanford HELM specialized suites data docs78.3%
39GPT 4osetting not named by the source· Stanford HELM specialized suites data docs76.7%
40Qwen2 72B Instruct· Stanford HELM specialized suites data docs75%
41Gemini 2.0 Flash-Lite Preview· Stanford HELM specialized suites data docs73.3%
42Gemini 2.5 Flash-Lite· Stanford HELM specialized suites data docs73.3%
43gemini-1.5-flash-002· Stanford HELM specialized suites data docs73.3%
44gemini-1.5-pro-002· Stanford HELM specialized suites data docs73.3%
45GLM 4.5 Air· Stanford HELM specialized suites data docs73.3%
46Llama 4 Scout· Stanford HELM specialized suites data docs73.3%
47Granite 4.0 H Small· Stanford HELM specialized suites data docs70%
48Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs70%
49GPT-4.1 nano· Stanford HELM specialized suites data docs68.3%
50Llama 3.1 8B Instruct Turbo· Stanford HELM specialized suites data docs66.7%

Showing 1–50 of 75 results.

Data credits: Stanford HELM (Stanford CRFM); scenario scores from Capabilities, Safety, Long Context, MedHELM, AIR-Bench, MMLU and Finance