Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - #3: Availability

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
1.26

Score distribution

71.1%
Mean
77.8%
Median
25.3%
Std dev
33.3%
10th percentile
100%
90th percentile
2.8%
Min
2.8%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.26 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
51Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs66.7%
52gemini-1.5-pro-002· Stanford HELM specialized suites data docs63.9%
53Qwen2 72B Instruct· Stanford HELM specialized suites data docs63.9%
54Grok 3· Stanford HELM specialized suites data docs59.7%
55GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs58.3%
56Yi 34B Chat· Stanford HELM specialized suites data docs58.3%
57Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs55.6%
58DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs54.2%
59Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs52.8%
60Qwen1.5-72B-Chat· Stanford HELM specialized suites data docs52.8%
61DeepSeek V3· Stanford HELM specialized suites data docs50%
62Llama 4 Scout· Stanford HELM specialized suites data docs50%
63Grok 3 Mini· Stanford HELM specialized suites data docs41.7%
64DeepSeek R1· Stanford HELM specialized suites data docs40.3%
65Mixtral 8x22B Instruct v0.1· Stanford HELM specialized suites data docs38.9%
66Grok 4· Stanford HELM specialized suites data docs36.1%
67Mistral Large2407· Stanford HELM specialized suites data docs36.1%
68O1 Mini· Stanford HELM specialized suites data docs33.3%
69GLM 4.5 Air· Stanford HELM specialized suites data docs22.2%
70Mixtral 8x7B Instruct v0.1· Stanford HELM specialized suites data docs22.2%
71Command R Plus· Stanford HELM specialized suites data docs19.4%
72Command R+· Stanford HELM specialized suites data docs19.4%
73Mistral Small· Stanford HELM specialized suites data docs13.9%
74Mistral (7B) Instruct v0.3· Stanford HELM specialized suites data docs6.9%
75DBRX-instruct· Stanford HELM specialized suites data docs2.8%

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