Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - Political Usage

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
1.01

Score distribution

62.9%
Mean
67.1%
Median
26.3%
Std dev
21.1%
10th percentile
94%
90th percentile
4.5%
Min
4.5%97.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.01 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-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs55.1%
52GPT-4o mini· Stanford HELM specialized suites data docs53.2%
53Llama 4 Scout· Stanford HELM specialized suites data docs50.5%
54GPT 4osetting not named by the source· Stanford HELM specialized suites data docs48.9%
55Yi 34B Chat· Stanford HELM specialized suites data docs48.4%
56DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs47.1%
57Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs44.5%
58GPT 4osetting not named by the source· Stanford HELM specialized suites data docs41.7%
59Qwen1.5-72B-Chat· Stanford HELM specialized suites data docs40.1%
60GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs39.9%
61Grok 3 Mini· Stanford HELM specialized suites data docs33.3%
62Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs31.6%
63DeepSeek V3· Stanford HELM specialized suites data docs30.8%
64GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs27.9%
65Mixtral 8x22B Instruct v0.1· Stanford HELM specialized suites data docs27.5%
66Grok 3· Stanford HELM specialized suites data docs24.9%
67O1 Mini· Stanford HELM specialized suites data docs23.7%
68Mixtral 8x7B Instruct v0.1· Stanford HELM specialized suites data docs21.1%
69Grok 4· Stanford HELM specialized suites data docs20.4%
70Command R+· Stanford HELM specialized suites data docs16.5%
71Mistral Large2407· Stanford HELM specialized suites data docs15.1%
72Command R Plus· Stanford HELM specialized suites data docs10.7%
73Mistral Small· Stanford HELM specialized suites data docs10.7%
74Mistral (7B) Instruct v0.3· Stanford HELM specialized suites data docs10%
75DBRX-instruct· Stanford HELM specialized suites data docs4.5%

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