SWE-Marathon v1.2: leaderboard
SWE-Marathon v1.2, the current release: binary task success over up to eight trials per task on 20 scored tasks (15 public sample and 5 held-out), grouped by model, agent and reasoning effort. Kept separate from the archived v1.0 and v1.1 task sets.
Metric: Pass@1 (%). Source: www.swe-marathon.org. Saturation forecast: Rough model projection: around 2026. 13 models tracked. Updated 21 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Claude Fable 5.2 | Thinking, Max | 86.79 | #1 |
| 2 | Claude Opus 5.5 | Thinking, Max | 81.25 | #9 |
| 3 | Claude Sonnet 5.5 | Thinking, Max | 80.63 | #24 |
| 4 | GPT-6.2 Astra | Thinking, Max | 77.5 | #4 |
| 5 | GPT-6 Astra | Thinking, Max | 68.75 | #10 |
| 6 | GPT-6.2 Sol | Thinking, Max | 66.88 | #11 |
| 7 | Gemini 4 Carbon | Thinking, high | 66.68 | #7 |
| 8 | Gemini 4 Neon | Thinking, xhigh | 62.5 | #3 |
| 9 | GPT-6.1 Sol | Thinking, Max | 60.38 | #15 |
| 10 | Gemini 4 Hydrogen | Thinking, xhigh | 58.13 | #16 |
| 11 | Gemini 4 Argon | Thinking, High | 54.46 | #19 |
| 12 | Claude Haiku 5.5 | Thinking, Max | 48.13 | #75 |
| 13 | Muse Wonder | Thinking, max | 38.13 | #35 |
Interactive version: theaggregate.ai/benchmark?slug=swe-marathon-v1-2 · How It Works · Data refreshed daily, snapshot 2026-11-21.