Skip to content

Benchmark

HELM Capabilities: WildBench - WB Score

People compare model answers and choose which they prefer. Scores are percentages. Higher is better.

Models measured
57
Scores recorded
59
Sources
1
IQ weight
1.45

Score distribution

79.8%
Mean
80.2%
Median
5.3%
Std dev
73.4%
10th percentile
85.7%
90th percentile
65.1%
Min
65.1%86.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.45 for chat models, measured over 57 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Qwen3 235B A22B Instruct 2507· Stanford HELM specialized suites data docs86.6%
2GPT 5.1· Stanford HELM specialized suites data docs86.3%
3Kimi K2 Instruct· Stanford HELM specialized suites data docs86.2%
4OpenAI o3· Stanford HELM specialized suites data docs86.1%
5Gemini 3 Pro Preview· Stanford HELM specialized suites data docs85.9%
6GPT 5· Stanford HELM specialized suites data docs85.7%
7Gemini 2.5 Pro Preview 03-25· Stanford HELM specialized suites data docs85.7%
8GPT-5 mini· Stanford HELM specialized suites data docs85.5%
9Claude Sonnet 4.5· Stanford HELM specialized suites data docs85.4%
10GPT 4.1· Stanford HELM specialized suites data docs85.4%
11o4 Mini· Stanford HELM specialized suites data docs85.4%
12Claude Opus 4thinking, 10k· Stanford HELM specialized suites data docs85.2%
13Grok 3· Stanford HELM specialized suites data docs84.9%
14GPT OSS 120B· Stanford HELM specialized suites data docs84.5%
15Claude Haiku 4.5· Stanford HELM specialized suites data docs83.9%
16Claude Sonnet 4thinking, 10k· Stanford HELM specialized suites data docs83.8%
17GPT-4.1 mini· Stanford HELM specialized suites data docs83.8%
18Claude Opus 4setting not named by the source· Stanford HELM specialized suites data docs83.3%
19DeepSeek V3· Stanford HELM specialized suites data docs83.1%
20GPT 4o· Stanford HELM specialized suites data docs82.8%
21DeepSeek R1· Stanford HELM specialized suites data docs82.8%
22Claude Sonnet 4setting not named by the source· Stanford HELM specialized suites data docs82.5%
23Gemini 2.5 Flash-Lite· Stanford HELM specialized suites data docs81.8%
24gemini-2.5-flash-preview-04-17· Stanford HELM specialized suites data docs81.7%
25Claude Sonnet 3.7· Stanford HELM specialized suites data docs81.4%
26gemini-1.5-pro-002· Stanford HELM specialized suites data docs81.3%
27GPT-4.1 nano· Stanford HELM specialized suites data docs81.1%
28Qwen3 Next 80B A3Bthinking· Stanford HELM specialized suites data docs80.7%
29GPT-5 nano· Stanford HELM specialized suites data docs80.6%
30Qwen2.5 72B Instruct Turbo· Stanford HELM specialized suites data docs80.2%
31Mistral Large2411· Stanford HELM specialized suites data docs80.1%
32Gemini 2.0 Flash· Stanford HELM specialized suites data docs80%
33Llama 4 Maverick· Stanford HELM specialized suites data docs80%
34Grok 4· Stanford HELM specialized suites data docs79.7%
35Claude 3.5 Sonnet· Stanford HELM specialized suites data docs79.2%
36gemini-1.5-flash-002· Stanford HELM specialized suites data docs79.2%
37GPT-4o mini· Stanford HELM specialized suites data docs79.1%
38Gemini 2.0 Flash-Lite Preview· Stanford HELM specialized suites data docs79%
39GLM 4.5 Air· Stanford HELM specialized suites data docs78.9%
40Mistral Small· Stanford HELM specialized suites data docs78.8%
41Nova Premier 1.0· Stanford HELM specialized suites data docs78.8%
42Llama 3.1 405B Instruct Turbo· Stanford HELM specialized suites data docs78.3%
43Palmyra X5· Stanford HELM specialized suites data docs78%
44Llama 4 Scout· Stanford HELM specialized suites data docs77.9%
45Nova Prov1· Stanford HELM specialized suites data docs77.7%
46Claude Haiku 3.5· Stanford HELM specialized suites data docs76%
47Meta Llama 3.1 70B Instruct Turbo· Stanford HELM specialized suites data docs75.8%
48Nova Litev1· Stanford HELM specialized suites data docs75%
49Nova Microv1· Stanford HELM specialized suites data docs74.3%
50Granite 3.3 8b Instruct· Stanford HELM specialized suites data docs74.1%

Showing 1–50 of 59 results.

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