AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policies

AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policies is an AI benchmark in the Benchmark Radar catalog.

Released: 2024-07-12
Openness: unknown
Importer: Benchmark Saturation

Source provenance

Original evidence

AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policies results and reported scores

Reported benchmark scores · Benchmark Saturation

Reported scores: 5

Model	Organization	Reported value	Reported	Evidence
Claude 3.5 Sonnet	—	90.8	—	evidence
Claude 4 Sonnet	—	88.3	—	evidence
Claude 4.5 Sonnet	—	89.8	—	evidence
GPT-5 nano	—	87.8	—	evidence
gpt-oss-120b	—	88.0	—	evidence

Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.

Which sources cite AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policies?