Skip to content

Benchmark

HELM AIR-Bench: AIRBench 2024 - #29: Unfair Market Practices

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
0.78

Score distribution

55.3%
Mean
55%
Median
34.1%
Std dev
10%
10th percentile
100%
90th percentile
3.3%
Min
3.3%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.78 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· Stanford HELM specialized suites data docs30%
52GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs30%
53Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs30%
54Yi 34B Chat· Stanford HELM specialized suites data docs28.3%
55Llama 4 Scout· Stanford HELM specialized suites data docs26.7%
56GPT-4o mini· Stanford HELM specialized suites data docs23.3%
57Grok 3 Mini· Stanford HELM specialized suites data docs23.3%
58GPT 4osetting not named by the source· Stanford HELM specialized suites data docs20%
59GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs18.3%
60DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs16.7%
61Grok 4· Stanford HELM specialized suites data docs16.7%
62Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs16.7%
63Grok 3· Stanford HELM specialized suites data docs15%
64DeepSeek R1· Stanford HELM specialized suites data docs13.3%
65Mixtral 8x22B Instruct v0.1· Stanford HELM specialized suites data docs13.3%
66Mixtral 8x7B Instruct v0.1· Stanford HELM specialized suites data docs13.3%
67GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs11.7%
68Mistral Large2407· Stanford HELM specialized suites data docs10%
69Qwen1.5-72B-Chat· Stanford HELM specialized suites data docs10%
70Command R Plus· Stanford HELM specialized suites data docs6.7%
71Command R+· Stanford HELM specialized suites data docs6.7%
72DeepSeek V3· Stanford HELM specialized suites data docs6.7%
73Mistral (7B) Instruct v0.3· Stanford HELM specialized suites data docs6.7%
74Mistral Small· Stanford HELM specialized suites data docs6.7%
75DBRX-instruct· Stanford HELM specialized suites data docs3.3%

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