Open-Sourced Recursive Self-Improvement Agent Beats GPT-5.5 on a Single RTX 4090
mark_k · x · 2026-08-02
Frontis-MA1 (35B) is a new AI4AI agent trained for recursive self-improvement in machine learning engineering. The project releases the full OpenMLE stack, closing the loop between learning and evolution using four post-trained operators: Draft, Improve, Debug, and Crossover.
Under a 12-hour per-task budget on a single RTX 4090 (capped at 12GB VRAM), the model improves its score on MLE-Bench Lite from a base 39.4% to 71.2% with OpenMLE-Evo-Max, beating GPT-5.5 + Codex. The gains also transfer to the held-out NatureBench. Weights and code are fully open-source.
Related event: OpenMLE: An Open-Source Stack for Recursive AI Self-Improvement(3 posts)→
More from coding & agent
- Reflection's Beam, a 501B-parameter open MoE model with 1M context, lands on Hugging Face under Apache 2.0 — QGallouedec · 2026-10-16
- 'LLM code is sloppy' rebutted with real ffmpeg code: identical if/else branches in a top library — kingslyj · 2026-10-16
- Office-editing tool calls need file, worktree and unit: lessons from dsh-univer-office — FunDistribution6537 · 2026-10-16
- Architect MD workflow: Gemini one-shot outputs that beat Opus, via 5-lens plan critique — oikk01 · 2026-10-16
- Prove a coding agent actually ran the tests: check artifacts, not approvals — ComprehensiveShake76 · 2026-10-16
- Building an acceptance harness for AI spreadsheet edits: allowlists, recalc checks, idempotency — ukdicha-modakkk · 2026-10-16