MazeBench is fully open source: any model with any harness can enter
xeophon · x · 2026-08-21
Responding to concerns that leaderboards only test frontier models with default harnesses, @xeophon clarified that MazeBench's code and all levels are fully open source—anyone can enter with any model and any harness, and the team will help you get it running. The author would love to see MazeBench conquered but doesn't think it will be solved any time soon.
Related event: MazeBench: Open-Source 3D Benchmark for Long-Horizon Agents(3 posts)→
More from Research
- MIT PhD Sharut Gupta joins DeepMind; Isola hails her representational alignment defense talk — phillip_isola · 2026-10-16
- 36 primitives, just 10 parameters each: a remarkably precise T-pose body fit — letandrewcook · 2026-10-16
- Interp researcher pushes back: if you can't simulate a neuron, you haven't explained it — nabla_theta · 2026-10-16
- MLP-Splatting (ECCV 2026) code released: MLPs as light-field primitives instead of Gaussians — AjdDavison · 2026-10-16
- Four-page solution to subcubic APSP builds intuition for Anthropic-AI-aided breakthrough — gautamcgoel · 2026-10-16
- ATAS: Zhejiang-led team builds world-state memory framework for consistent agentic video generation — CSProfKGD · 2026-10-16