Developer Runs 1,200+ Blender Modeling Sessions to Blind-Rank LLMs and Agent Harnesses
Izolight · reddit · 2026-10-06
Developer Izolight launched Render Arena, a blind A/B arena where AI agents build models in Blender and viewers vote on results.
- 1,200+ runs comparing agent harnesses: pi, opencode, omp, codex, Claude Code, and dsh
- Thesis: most arenas test models, not harnesses; also contrasts agents writing scripts directly into Blender vs. via MCP
- Probed reasoning levels to find cost/time sweet spots
- New second mode: reconstruct a reference image via 3D modeling; users can submit their own runs to fill gaps
- Inspired by minebench.ai, with initial prompts borrowed from it; rankings depend on community votes
More from coding & agent
- 462GB DeepSeek model runs on two desk-side DGX Sparks with experts squeezed to 2.77 bits — Teknium · 2026-10-06
- Dev pitches 'Business in a Box': ship a whole AI agent company inside one Claude Code account — RileyRalmuto · 2026-10-06
- Armin Ronacher: a Pi release took just 4 hours thanks to GitHub Actions — mitsuhiko · 2026-10-06
- Building a Proactive Personal Agent With Pi Durable: Checkpoints, Memory, Own Linux Desktop — omarsar0 · 2026-10-06
- ThePrimeagen's Desktop QA Agent: Cut QEMU Runs From 15 Minutes to 2, Still Too Slow — cyrus_zei · 2026-10-06
- Mantine 9.7 Ships: MCP Server Now Indexes 406 Doc Pages, Cuts Payloads ~90% — DavidWells · 2026-10-06