# IBM and Kaggle open maintained leaderboard for Java migration agents

_Published Tuesday, October 27, 2026 at 8:11 AM EDT · AI · Latest · Tier 2 — Notable_

IBM and Kaggle have launched a maintained ScarfBench leaderboard evaluating whether coding agents can migrate enterprise Java applications while preserving their behavior. IBM says Kaggle independently reproduced the evaluation on its platform, sponsored the compute and will add new models over time.

Across 204 migration tasks, Gemini 4 Argon with Antigravity led the weighted smoke-test metric at 75.2%. Claude Opus 5.5 with Claude Code had the highest all-tests-pass rate at 72.1%. Results measure models paired with agent software, rather than models alone. Of 745 classified failing runs, 535 involved runtime behavioral failures. Public model submission is not available at launch.

## Sources

- [IBM Research](https://research.ibm.com/blog/ibm-and-kaggle-launch-leaderboard-for-java-framework-migration)

---
Canonical: https://techandbusiness.org/newswire/ev6JamuCkgjQQg85BHtHym
Published: 2026-10-27T12:11:24.007Z
Story chronology: 2026-10-27T12:00:00.000Z
Retrieved: 2026-10-27T14:40:48.861Z
Publisher: Tech & Business (techandbusiness.org)