HELM · WildBench
The Frontier
Best score over time · one chart, every benchmark
Distribution
Where models cluster
Correlated benchmarks
Pearson r · original research
Benchmarks that track with HELM · WildBench
Pearson correlation across models scored on both benchmarks. Closer to 1 = strongly predictive.
Full rankings
34 models tested · sorted by score
| # | Model | Score |
|---|---|---|
| 1 |  GPT-5.1 | 86.3 |
| 2 |  Kimi K2 0711 | 86.2 |
| 3 | o3 | 86.1 |
| 4 |  Gemini 3 Pro | 85.9 |
| 5 | Gemini 2.5 Pro | 85.7 |
| 6 | GPT-5 Chat | 85.7 |
| 7 | GPT-5 Mini | 85.5 |
| 8 | GPT-4.1 | 85.4 |
| 9 | o4 Mini | 85.4 |
| 10 |  Grok 3 Beta | 84.9 |
| 11 | gpt-oss-120b | 84.5 |
| 12 | GPT-4.1 Mini | 83.8 |
| 13 |  DeepSeek V3 | 83.1 |
| 14 | GPT-4o (2024-11-20) | 82.8 |
| 15 | R1 0528 | 82.8 |
| 16 | Gemini 2.5 Flash Lite | 81.8 |
| 17 | Gemini 2.5 Flash | 81.7 |
| 18 |  Claude 3.7 Sonnet | 81.4 |
| 19 | Gemini 1.5 Pro (Feb 2024) | 81.3 |
| 20 | GPT-4.1 Nano | 81.1 |
| 21 |  Qwen3 Next 80B A3B Thinking | 80.7 |
| 22 | GPT-5 Nano | 80.6 |
| 23 |  Mistral Large 2411 | 80.1 |
| 24 | Gemini 2.0 Flash | 80.0 |
| 25 | Grok 4 | 79.7 |
| 26 | Claude 3.5 Sonnet | 79.2 |
| 27 | Gemini 1.5 Flash (May 2024) | 79.2 |
| 28 | GPT-4o-mini (2024-07-18) | 79.1 |
| 29 | Gemini 2.0 Flash Lite | 79.0 |
| 30 | Mistral Small 3.1 24B | 78.8 |
| 31 |  Palmyra X5 | 78.0 |
| 32 | Claude 3.5 Haiku | 76.0 |
| 33 | gpt-oss-20b | 73.7 |
| 34 | Grok 3 Mini Beta | 65.1 |
Frequently asked
Pulled from the HELM · WildBench dataset · updated daily
What does HELM · WildBench measure?
HELM · WildBench is a knowledge benchmark in the BenchGecko catalog. 34 AI models have been tested on it. Scores range from 65.1 to 86.3 out of 100.
Which model leads on HELM · WildBench?
GPT-5.1 from OpenAI leads HELM · WildBench with a score of 86.3. The median score across 34 tested models is 81.6.
Is HELM · WildBench saturated?
No · the top score is 86.3 out of 100 (86%). There is still meaningful room for improvement on HELM · WildBench.
Does HELM · WildBench predict performance on other benchmarks?
Yes · HELM · WildBench scores correlate 0.96 with Artificial Analysis · Coding Index across 5 shared models. Models that do well on HELM · WildBench tend to do well on Artificial Analysis · Coding Index.
How often is HELM · WildBench data refreshed?
BenchGecko pulls updates daily. New model scores on HELM · WildBench appear as soon as they are published by Epoch AI or the model provider.
Top on HELM · WildBench
GPT-5.1 · 86.3Kimi K2 0711 · 86.2o3 · 86.1Gemini 3 Pro · 85.9Gemini 2.5 Pro · 85.7More knowledge benchmarks
Same category · related evaluations