ScarfBench (Kaggle) - Deploy: leaderboard
Coding agents (one model at one effort inside its harness, run by Kaggle) carry out 204 migrations of enterprise Java applications between Jakarta EE, Quarkus and Spring; the score is the share of the 204 migrations whose application builds and deploys, the gate before any smoke test runs. Unlike the ScarfBench paper board (174 focused-app migrations, one skills prompt), Kaggle runs and maintains these results over focused and whole applications.
Metric: Migrations that build and deploy (%) of 204 Java framework migrations, each model run in a coding-agent harness by Kaggle. Source: www.kaggle.com. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon (High) | 97.55 | #11 (Gemini 4 Argon) |
| 2 | Claude Opus 5 (xHigh) | 97.55 | #22 (Claude Opus 5) |
| 3 | Gemini 3.8 Flash (Medium) | 97.06 | #36 (Gemini 3.8 Flash) |
| 4 | Kimi K3 (Max) | 96.08 | #54 (Kimi K3) |
| 5 | GPT-6 Astra (High) | 96.08 | #6 (GPT-6 Astra) |
| 6 | GPT-6.1 Sol (Max) | 95.59 | #10 (GPT-6.1 Sol) |
| 7 | Claude Opus 5.5 (Max) | 95.1 | #5 (Claude Opus 5.5) |
| 8 | DeepSeek V4.1 Flash (Max) | 94.61 | #59 (DeepSeek V4.1 Flash) |
| 9 | Qwen 3.8 Max (xHigh) | 94.12 | #55 (Qwen 3.8 Max) |
| 10 | Gemini 3.1 Pro (Preview) (High) | 83.33 | #94 (Gemini 3.1 Pro (Preview)) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=scarfbench-kaggle-deploy · How It Works · Data refreshed daily, snapshot 2026-11-13.