Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - #7: Supporting Malicious Organized Groups

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
1.07

Score distribution

66.3%
Mean
68.9%
Median
29.6%
Std dev
17.8%
10th percentile
97.8%
90th percentile
2.2%
Min
2.2%98.9%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.07 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 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs98.9%
2Claude 3.5 Sonnetsetting not named by the source· Stanford HELM specialized suites data docs98.9%
3Claude Haiku 4.5· Stanford HELM specialized suites data docs98.9%
4GPT OSS 20b· Stanford HELM specialized suites data docs98.9%
5Claude 3 Sonnet· Stanford HELM specialized suites data docs97.8%
6Claude Sonnet 4.5· Stanford HELM specialized suites data docs97.8%
7GPT OSS 120B· Stanford HELM specialized suites data docs97.8%
8Granite 3.3 8b Instruct· Stanford HELM specialized suites data docs97.8%
9Llama3 8B Chat· Stanford HELM specialized suites data docs97.8%
10OpenAI o3· Stanford HELM specialized suites data docs97.8%
11GPT 5· Stanford HELM specialized suites data docs96.7%
12o4 Mini· Stanford HELM specialized suites data docs96.7%
13Qwen3 Next 80B A3Bthinking· Stanford HELM specialized suites data docs96.7%
14Claude 3 Opus· Stanford HELM specialized suites data docs95.6%
15Claude Opus 4· Stanford HELM specialized suites data docs95.6%
16Claude Sonnet 4· Stanford HELM specialized suites data docs95.6%
17GPT-5 mini· Stanford HELM specialized suites data docs95.6%
18GPT-5 nano· Stanford HELM specialized suites data docs95.6%
19Claude 3 Haiku· Stanford HELM specialized suites data docs94.4%
20o1· Stanford HELM specialized suites data docs94.4%
21Palmyra X5· Stanford HELM specialized suites data docs94.4%
22GPT 5.1· Stanford HELM specialized suites data docs93.3%
23Qwen3 235B A22B Instruct 2507· Stanford HELM specialized suites data docs92.2%
24Llama 3.1 8B Instruct Turbo· Stanford HELM specialized suites data docs91.1%
25Gemini 2.5 Pro Preview 03-25· Stanford HELM specialized suites data docs88.9%
26Granite 4.0 H Small· Stanford HELM specialized suites data docs88.9%
27o3 Mini· Stanford HELM specialized suites data docs88.9%
28Claude Sonnet 3.7· Stanford HELM specialized suites data docs87.8%
29GPT-4.5 Preview· Stanford HELM specialized suites data docs87.8%
30Kimi K2 Instruct· Stanford HELM specialized suites data docs83.3%
31Llama 4 Maverick· Stanford HELM specialized suites data docs82.2%
32GPT-4 Turbo· Stanford HELM specialized suites data docs80%
33GPT 4.1· Stanford HELM specialized suites data docs77.8%
34GPT 4· Stanford HELM specialized suites data docs76.7%
35Gemini 2.0 Flash-Lite Preview· Stanford HELM specialized suites data docs75.6%
36gemini-2.5-flash-preview-04-17· Stanford HELM specialized suites data docs74.4%
37GPT-4.1 nano· Stanford HELM specialized suites data docs74.4%
38Gemini 2.5 Flash-Lite· Stanford HELM specialized suites data docs68.9%
39Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs68.9%
40GPT-4o mini· Stanford HELM specialized suites data docs67.8%
41GPT-4.1 mini· Stanford HELM specialized suites data docs66.7%
42gemini-2.0-pro-exp-02-05· Stanford HELM specialized suites data docs65.6%
43GPT 4osetting not named by the source· Stanford HELM specialized suites data docs65.6%
44GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs64.4%
45Gemini 2.0 Flash· Stanford HELM specialized suites data docs63.3%
46GLM 4.5 Air· Stanford HELM specialized suites data docs63.3%
47GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs63.3%
48Gemini 3 Pro Preview· Stanford HELM specialized suites data docs62.2%
49Qwen2 72B Instruct· Stanford HELM specialized suites data docs62.2%
50Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs62.2%

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