AutoBaxBench - Hard - Correct: leaderboard
AutoBaxBench is the automatically generated extension of BaxBench, built to limit benchmark contamination; Hard tasks require implementing five API endpoints. This board is pass@1, the percentage of generated backends that pass the functional tests regardless of security; higher is better.
Metric: Correct solutions, pass@1 (%). Source: baxbench.com. Saturation forecast: Estimated already saturated. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 82.86 | #148 |
| 2 | Gemini 3 Pro | 76.17 | #82 |
| 3 | Claude Sonnet 4 | 68.57 | #207 |
| 4 | Claude 3.7 Sonnet | 64.76 | #259 |
| 5 | GPT-5 | 59.29 | #105 |
| 6 | Grok 4 | 42.14 | #189 |
| 7 | DeepSeek R1 | 35 | #262 |
| 8 | Gemini 2.5 Pro | 31.43 | #159 |
| 9 | GPT-4o | 10.71 | #357 |
| 10 | Llama 3.3 70B | 2.86 | #578 |
| 11 | Qwen 2.5 72B | 2.38 | #512 |
| 12 | Qwen 2.5 7B | 0.24 | #895 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=autobaxbench-hard-correct · How It Works · Data refreshed daily, snapshot 2026-11-05.