Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - #44: Illegal services/exploitation

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
1.51

Score distribution

72%
Mean
82.2%
Median
26.5%
Std dev
31.1%
10th percentile
98.9%
90th percentile
2.2%
Min
2.2%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.51 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
51GPT-4.1 mini· Stanford HELM specialized suites data docs66.7%
52Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs66.7%
53GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs64.4%
54GPT 4osetting not named by the source· Stanford HELM specialized suites data docs62.2%
55Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs62.2%
56Llama 4 Scout· Stanford HELM specialized suites data docs60%
57Qwen1.5-72B-Chat· Stanford HELM specialized suites data docs51.1%
58Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs51.1%
59DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs48.9%
60Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs48.9%
61Grok 3· Stanford HELM specialized suites data docs46.7%
62Grok 4· Stanford HELM specialized suites data docs46.7%
63Grok 3 Mini· Stanford HELM specialized suites data docs44.4%
64GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs42.2%
65DeepSeek R1· Stanford HELM specialized suites data docs41.1%
66DeepSeek V3· Stanford HELM specialized suites data docs37.8%
67Mixtral 8x22B Instruct v0.1· Stanford HELM specialized suites data docs34.4%
68O1 Mini· Stanford HELM specialized suites data docs31.1%
69Mistral Large2407· Stanford HELM specialized suites data docs22.2%
70Command R+· Stanford HELM specialized suites data docs20%
71Mixtral 8x7B Instruct v0.1· Stanford HELM specialized suites data docs20%
72Command R Plus· Stanford HELM specialized suites data docs15.6%
73Mistral Small· Stanford HELM specialized suites data docs15.6%
74Mistral (7B) Instruct v0.3· Stanford HELM specialized suites data docs4.4%
75DBRX-instruct· Stanford HELM specialized suites data docs2.2%

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