Today’s best coding agents rely on massive API models running in the cloud. What if every developer could rely on an autonomous agent equally capable offline, on consumer hardware?
Post-train an open model into a reliable agent that navigates complex codebases and drafts fixes for real software issues, accelerating developer workflows on everyday hardware.
Today’s most capable coding agents rely on internet-connected models running on powerful commercial infrastructure. That puts advanced software engineering out of reach for many developers.
Small local models have made rapid progress, but they still struggle with complex, multi-turn software engineering. Large codebases are hard to navigate, making reliable bug fixes and feature additions tough for models that fit on a single accelerator.
While The Gemma 4 Developer Agent Competition leaderboard ranks participants by bugs fixed, the Paper Track invites you to contribute research to advance the state-of-the-art in agentic software engineering. You will submit a Kaggle Writeup that describes your methods, experiments, and other details. Participation in the main competition isn’t required to submit to the Paper Track, though we encourage participants to join both. Submissions can include any related research, as long as it is unpublished.
Your solution could help put powerful coding assistants in the hands of millions more developers. Sharing how you did it makes advanced coding tools accessible to everyone.
Awards:- $35,000
Deadline:- 12-11-2026
