LongCat 2.5 Release - SWE Atlas - Test Writing: leaderboard
LongCat-2.5 final release detailed native result table, November 2026. External:true cells are explicitly cited other-evaluator results and excluded from this component; self-run numeric cells retained. No Preview/final or release-generation aliases. Detailed table used instead of inconsistent rounded headline charts. SWE Atlas - Codebase QnA and Test Writing: Claude Code (v 2.1.116) with Claude Opus 4.5 as the judge (majority vote over three judgments); results are averaged over three independent runs. Metric: Score (%).
Metric: Score (%). Source: longcat.ai. Saturation forecast: Around July 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 | 62.2 | #16 |
| 2 | Claude Opus 4.8 | 49.6 | #39 |
| 3 | DeepSeek V4.1 Flash | 48.9 | #37 |
| 4 | GPT-5.6 Sol | 45.9 | #22 |
| 5 | GPT-5.5 | 42.6 | #33 |
| 6 | Kimi K3 | 42.2 | #35 |
| 7 | GLM-5.3 | 42.2 | #44 |
| 8 | LongCat 2.0 | 28 | #220 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=longcat-2-5-release-swe-atlas-test-writing · How It Works · Data refreshed daily, snapshot 2026-11-05.