SEAL - SWE Atlas - Test Writing: leaderboard
Scale SEAL evaluation of LLM ability to write test suites for software engineering tasks.
Metric: Score. Source: scale.com. Status: years away from saturation. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Sol (xHigh) | 45.9 |
| 2 | GPT-5.5 (xHigh) | 42.59 |
| 3 | GLM-5.2 | 41.48 |
| 4 | GPT-5.4 (xHigh) | 40 |
| 5 | Muse Spark | 31.11 |
| 6 | Gemini 3 Flash | 30.3 |
| 7 | Gemini 3.1 Pro (Preview) | 29.84 |
| 8 | GLM-5 | 28.74 |
| 9 | DeepSeek V4 Pro | 27.05 |
| 10 | Kimi K2.5 | 25.77 |
| 11 | MiniMax-M2.5 | 18.6 |
Interactive version: theaggregate.ai/benchmark?slug=seal-swe-atlas-test-writing · How It Works · Data refreshed daily, snapshot 2026-09-05.