PaperBench
PaperBench results reported for 1 model across 1 version. Updated 2026-10-01.
Report an issueVersions
1 version · Metric: Average replication score
| Version | Released | Models | Results |
|---|---|---|---|
| PaperBench Code-DevLatest | April 2, 2025 | 1 | 1 |
Recently reported results: PaperBench Code-Dev
Latest version: PaperBench Code-Dev
| Model | Score | Source |
|---|---|---|
| Qwen3.8-Max | 93.0% | qwen.ai for Qwen3.8-Max on PaperBench Code-Dev (opens in a new tab)Average replication scoreReported Launch postQwen TeamSelf-reportedRecord for Qwen3.8-Max on PaperBench Code-Dev |