Skip to content

Benchmark

HELM Safety: Anthropic Red Team - LM Evaluated Safety score

Safety benchmarks examine risks in model behaviour, including how models respond to unsafe requests. Scores are percentages. Higher is better.

Models measured
69
Scores recorded
74
Sources
1
IQ weight
1.07

Score distribution

97.7%
Mean
99%
Median
4.4%
Std dev
95.7%
10th percentile
99.8%
90th percentile
71.1%
Min
71.1%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.07 for chat models, measured over 69 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude 3 Haiku· Stanford HELM specialized suites data docs100%
2gemini-1.5-flash-001· Stanford HELM specialized suites data docs99.9%
3gemini-1.5-pro-001· Stanford HELM specialized suites data docs99.9%
4Qwen3 Next 80B A3Bthinking· Stanford HELM specialized suites data docs99.9%
5Qwen3 235B A22B Instruct 2507· Stanford HELM specialized suites data docs99.9%
6Claude 3 Sonnet· Stanford HELM specialized suites data docs99.8%
7Claude 3 Opus· Stanford HELM specialized suites data docs99.8%
8Claude 3.5 Sonnet· Stanford HELM specialized suites data docs99.8%
9GPT-4 Turbo· Stanford HELM specialized suites data docs99.7%
10Claude Sonnet 3.7· Stanford HELM specialized suites data docs99.7%
11GPT OSS 20b· Stanford HELM specialized suites data docs99.7%
12GPT-4.1 nano· Stanford HELM specialized suites data docs99.6%
13Granite 4.0 H Small· Stanford HELM specialized suites data docs99.6%
14GPT-3.5 Turbosetting not named by the source· Stanford HELM specialized suites data docs99.6%
15Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs99.6%
16GPT-5 nano· Stanford HELM specialized suites data docs99.6%
17Grok 3 Mini· Stanford HELM specialized suites data docs99.5%
18GPT 5.1· Stanford HELM specialized suites data docs99.5%
19Gemini 2.5 Pro Preview 03-25· Stanford HELM specialized suites data docs99.5%
20GPT OSS 120B· Stanford HELM specialized suites data docs99.5%
21DeepSeek LLM 67B Chat· Stanford HELM specialized suites data docs99.4%
22Gemini 2.0 Flash· Stanford HELM specialized suites data docs99.4%
23GPT-4.5 Preview· Stanford HELM specialized suites data docs99.4%
24Gemini 2.0 Flash-Lite Preview· Stanford HELM specialized suites data docs99.4%
25Olmo 2 32B Instruct· Stanford HELM specialized suites data docs99.3%
26GPT-4.1 mini· Stanford HELM specialized suites data docs99.3%
27GPT 4.1· Stanford HELM specialized suites data docs99.3%
28Kimi K2 Instruct· Stanford HELM specialized suites data docs99.3%
29Palmyra X5· Stanford HELM specialized suites data docs99.3%
30Claude Sonnet 4thinking, 10k· Stanford HELM specialized suites data docs99.2%
31GPT 4o· Stanford HELM specialized suites data docs99.1%
32Granite 3.3 8b Instruct· Stanford HELM specialized suites data docs99.1%
33GPT-5 mini· Stanford HELM specialized suites data docs99.1%
34GPT 5· Stanford HELM specialized suites data docs99.1%
35Qwen2 72B Instruct· Stanford HELM specialized suites data docs99.1%
36DeepSeek R1setting not named by the source· Stanford HELM specialized suites data docs99.1%
37GLM 4.5 Air· Stanford HELM specialized suites data docs99%
38Qwen1.5-72B-Chat· Stanford HELM specialized suites data docs99%
39Claude Opus 4setting not named by the source· Stanford HELM specialized suites data docs98.9%
40Claude Opus 4thinking, 10k· Stanford HELM specialized suites data docs98.9%
41Llama3 8B Chat· Stanford HELM specialized suites data docs98.8%
42gemini-2.5-flash-preview-04-17· Stanford HELM specialized suites data docs98.8%
43gemini-2.0-pro-exp-02-05· Stanford HELM specialized suites data docs98.7%
44Gemini 2.5 Flash-Lite· Stanford HELM specialized suites data docs98.7%
45o3 Mini· Stanford HELM specialized suites data docs98.6%
46Qwen2.5 7B Instruct Turbo· Stanford HELM specialized suites data docs98.5%
47Claude Sonnet 4setting not named by the source· Stanford HELM specialized suites data docs98.4%
48GPT-4o mini· Stanford HELM specialized suites data docs98.3%
49o1· Stanford HELM specialized suites data docs98.3%
50O1 Mini· Stanford HELM specialized suites data docs98.3%

Showing 1–50 of 74 results.

Data credits: Stanford HELM (Stanford CRFM); scenario scores from Capabilities, Safety, Long Context, MedHELM, AIR-Bench, MMLU and Finance