Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - #10: Depicting Violence

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
0.50

Score distribution

43.9%
Mean
38%
Median
17.6%
Std dev
24.7%
10th percentile
70%
90th percentile
18%
Min
18%90.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.50 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
51DBRX-instruct· Stanford HELM specialized suites data docs33.3%
52GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs32.7%
53GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs32%
54Mistral Small· Stanford HELM specialized suites data docs32%
55GPT-4.1 mini· Stanford HELM specialized suites data docs30%
56GPT-4o mini· Stanford HELM specialized suites data docs30%
57Gemini 3 Pro Preview· Stanford HELM specialized suites data docs28.7%
58Command R Plus· Stanford HELM specialized suites data docs28%
59GPT 4osetting not named by the source· Stanford HELM specialized suites data docs28%
60Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs28%
61O1 Mini· Stanford HELM specialized suites data docs28%
62gemini-1.5-flash-002· Stanford HELM specialized suites data docs27.3%
63Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs27.3%
64o3 Mini· Stanford HELM specialized suites data docs27.3%
65Command R+· Stanford HELM specialized suites data docs26.7%
66GPT 4.1· Stanford HELM specialized suites data docs26.7%
67Gemini 2.5 Pro Preview 03-25· Stanford HELM specialized suites data docs25.3%
68gemini-2.5-flash-preview-04-17· Stanford HELM specialized suites data docs24.7%
69DeepSeek V3· Stanford HELM specialized suites data docs24%
70Grok 3· Stanford HELM specialized suites data docs24%
71Grok 3 Mini· Stanford HELM specialized suites data docs24%
72DeepSeek R1· Stanford HELM specialized suites data docs23.3%
73gemini-1.5-pro-002· Stanford HELM specialized suites data docs22.7%
74GLM 4.5 Air· Stanford HELM specialized suites data docs18.7%
75Grok 4· Stanford HELM specialized suites data docs18%

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