Dev gives System 1 decision models virtual gamepads in a Godot AI fighting tournament
ViceGodIm · reddit · 2026-10-28
A developer built a Godot sword-fighting tournament for eight local entrants, mostly System 1-style decision models plus a GLiNER classifier wildcard. The loop: structured game state → legal action menu → model choice → short gamepad input, with inference running asynchronously alongside the 60Hz game; held inputs can chain attacks. The biggest snag was action selection — one model always chose guard via argmax until switched to probability sampling. Honest caveats: response times differ across models and each matchup ran only once. Demo video available.
More from coding & agent
- Anthropic models burn 2-3x tokens and cost 10-50x more on FrontierCode 2.0 — teortaxesTex · 2026-10-28
- Devs praise OpenAI's Decisions API: Codex integration drives full-time switch — craigsdennis · 2026-10-28
- Personal AI agent trains a chess transformer on dual RTX 4090 and battles Sunfish — andreisavu · 2026-10-28
- FrontierCode 2: Anthropic models burn 2-3x tokens, cost 10-50x more than open rivals — bookwormengr · 2026-10-28
- Opus 5.5 ultracode workflow with 9 subagents builds a Three.js scene — nptacek · 2026-10-28
- Agent sandboxes talk at PLATFORMday SF: the missing runtime for AI — gethackteam · 2026-10-28