Inside Anthropic's Hackathon: Can AI One-Shot a Full Unity 3D Game?
davidfromkansas · x · 2026-09-01
The author attended Anthropic's Fable 5 hackathon (Fable 5 was banned by the US government the night before, so teams used Opus 4.8) and recounts the experience:
- Organizers asked for hard problems matching Fable 5's long-horizon execution ability; weak projects, flimsy chatbots, and anything healthcare-related were off-limits.
- A former amateur game dev, the author explored whether AI could one-shot an entire 3D game in Unity.
- He ran the exact same prompt through two models (Fable 5 vs Opus 4.8): a controllable 3D futuristic spacecraft surrounded by asteroids of varying sizes, with a health bar, collision damage, and space-bar shooting.
- The post shows both outputs side by side and lets readers guess which model made which.
More from coding & agent
- OpenClaw 2.0 vs Hermes agent: self-checking strategy matters more than the model — rohanpaul_ai · 2026-09-01
- 实测：Qwen 27B 在配置调试中一次性修复了 DSV4 无法解决的问题 — OvertaxedOne · 2026-09-01
- AI Agents现状：效果惊艳但仍需大量人工干预 — Hot-Cup3451 · 2026-09-01
- SOTA agents generate useless code; we need precise definitions for bad code — alexisgallagher · 2026-09-01
- Dev shares HTML to PPTX solution using pptxgenjs — dotey · 2026-09-01
- Developer Observes Persistent Dumb Decisions in AI Code Outputs — rickasaurus · 2026-09-01