Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - #30: Disempowering Workers

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
1.03

Score distribution

64.2%
Mean
66.7%
Median
27%
Std dev
26.7%
10th percentile
96.7%
90th percentile
7.5%
Min
7.5%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.03 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 4osetting not named by the source· Stanford HELM specialized suites data docs50%
52GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs50%
53Yi 34B Chat· Stanford HELM specialized suites data docs49.2%
54GPT 4· Stanford HELM specialized suites data docs45%
55GPT-4o mini· Stanford HELM specialized suites data docs43.3%
56DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs41.7%
57Mixtral 8x22B Instruct v0.1· Stanford HELM specialized suites data docs41.7%
58DeepSeek R1· Stanford HELM specialized suites data docs37.5%
59GPT 4osetting not named by the source· Stanford HELM specialized suites data docs36.7%
60Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs35.8%
61Qwen1.5-72B-Chat· Stanford HELM specialized suites data docs35%
62GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs33.3%
63O1 Mini· Stanford HELM specialized suites data docs33.3%
64Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs30%
65Mixtral 8x7B Instruct v0.1· Stanford HELM specialized suites data docs30%
66Grok 3· Stanford HELM specialized suites data docs27.5%
67Grok 4· Stanford HELM specialized suites data docs27.5%
68GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs26.7%
69DeepSeek V3· Stanford HELM specialized suites data docs24.2%
70Mistral (7B) Instruct v0.3· Stanford HELM specialized suites data docs20%
71Mistral Large2407· Stanford HELM specialized suites data docs20%
72Command R+· Stanford HELM specialized suites data docs18.3%
73Mistral Small· Stanford HELM specialized suites data docs14.2%
74Command R Plus· Stanford HELM specialized suites data docs13.3%
75DBRX-instruct· Stanford HELM specialized suites data docs7.5%

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