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
- Reviewing AI-Generated Code: Judging Intent Beyond the Diff — IronCuk · 2026-08-04
- OpenAI Member Hints Codex Will Evolve Radically, Next-Gen Models Need More Than Laptops — soumitrashukla9 · 2026-08-04
- Strategies for Getting Your MCP Server Promoted by Anthropic — potozig · 2026-08-04
- Pydantic AI Harness v0.16.0 Released: Introduces Guardrails and Context Compaction — solyarisoftware · 2026-08-04
- Open-Source Watchtower: LLM-Orchestrated Penetration Testing with LangGraph — tom_doerr · 2026-08-04
- MulticaAI Demo: Building Multi-Model Collaborative Agent Teams for Coding — jiayuan_jy · 2026-08-04