Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - Operational Misuses

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
0.80

Score distribution

37.1%
Mean
34.5%
Median
15.6%
Std dev
17.1%
10th percentile
59%
90th percentile
10.9%
Min
10.9%75.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.80 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
51Grok 3 Mini· Stanford HELM specialized suites data docs26.8%
52Kimi K2 Instruct· Stanford HELM specialized suites data docs26.7%
53GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs26.4%
54GPT-4.1 mini· Stanford HELM specialized suites data docs26%
55GPT-4o mini· Stanford HELM specialized suites data docs25.6%
56GLM 4.5 Air· Stanford HELM specialized suites data docs25.4%
57Qwen2 72B Instruct· Stanford HELM specialized suites data docs25.4%
58Llama 4 Scout· Stanford HELM specialized suites data docs24.5%
59GPT 4.1· Stanford HELM specialized suites data docs23.5%
60GPT-4.1 nano· Stanford HELM specialized suites data docs22.6%
61Grok 4· Stanford HELM specialized suites data docs21.9%
62Mixtral 8x7B Instruct v0.1· Stanford HELM specialized suites data docs20.9%
63Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs20.4%
64Grok 3· Stanford HELM specialized suites data docs20%
65GPT 4osetting not named by the source· Stanford HELM specialized suites data docs19.4%
66Mistral (7B) Instruct v0.3· Stanford HELM specialized suites data docs19.4%
67Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs17.8%
68DBRX-instruct· Stanford HELM specialized suites data docs17.1%
69DeepSeek R1· Stanford HELM specialized suites data docs16.3%
70Mistral Large2407· Stanford HELM specialized suites data docs15%
71Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs14.9%
72Command R+· Stanford HELM specialized suites data docs14.4%
73DeepSeek V3· Stanford HELM specialized suites data docs12.4%
74Command R Plus· Stanford HELM specialized suites data docs11.8%
75Mistral Small· Stanford HELM specialized suites data docs10.9%

Showing 51–75 of 75 results.

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