APEX-SWE: leaderboard
Integration and observability engineering tasks in containerized repositories, checked by program; the score is the mean over 4 attempts on Mercor's 200-task held-out set, supplied by Mercor. It targets service wiring and monitoring work rather than issue fixing.
Metric: Mean score over 4 attempts of 200 tasks (%). Source: fireworks.ai. Saturation forecast: Around 2029. 10 models tracked. Updated 16 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Max | 67.62 | #5 |
| 2 | Claude Opus 5 | Max | 63.75 | #20 |
| 3 | Grok 4.6 | xHigh | 57.75 | #45 |
| 4 | Grok 4.7 | xHigh | 53.62 | #47 |
| 5 | DeepSeek V4.1 Flash | Max | 51.88 | #62 |
| 6 | GPT-6 Astra | Max | 50 | #6 |
| 7 | DeepSeek V4 Pro (0813) | Max | 46.92 | #90 |
| 8 | GPT-6.1 Sol | Max | 46.88 | #10 |
| 9 | GLM-5.3 | Max | 45.62 | #68 |
| 10 | Gemini 3.8 Flash | High | 36.62 | #36 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=apex-swe · How It Works · Data refreshed daily, snapshot 2026-11-16.