Matt Shumer burns 400B tokens testing Fable 5.2, shows off 5 wild projects
FinanceYF5 · x · 2026-11-10
Matt Shumer skipped the benchmarks and tested Fable 5.2 by having it actually build things, using over 400 billion tokens. His five standout demos:
- Voxel destruction engine from one prompt: buildings and objects shatter in real time with full physics and sound feedback — and it runs on a MacBook;
- A 4-minute-44-second video generated in one pass: fully autonomous, no manual clip stitching. His verdict: "it makes Opus look like a toy";
- A "mirror Earth" open-world game: travel anywhere, destroy anything, with weather synced to the real planet;
- A "Harry Potter × Fortnite" battle royale from a single mashup prompt, playable in the browser;
- A robot the model designed for assembly: built around real manufacturing constraints in his apartment.
The throughline: shifting evaluation from scores to end-to-end project delivery — complete works generated in one shot, spanning games, video, physics engines, and hardware design.
Related event: HyperWrite founder tests Fable 5.2 with 400B tokens, unveils wildest builds(7 posts)→
More from Models
- Google's Logan Kilpatrick teases with a one-word 'Gemini' reply, hinting at upcoming launch — ammaar · 2026-11-10
- Meta Launches Muse Ember: Its Smallest Model Sits on the Efficiency Frontier — Simon Willison · 2026-11-10
- Kimi K3.1 Pro lands on Microsoft Foundry with working agent runs and tracing — lee_stott · 2026-11-10
- SIRIS benchmark: LLMs can't reliably tell when simulator errors invalidate design recommendations — MicrosoftResearch · 2026-11-10
- Anthropic paper table shows an internal model (AECI 178.78) beating all released Claude models — kimmonismus · 2026-11-10
- OpenAI: ChatGPT Search Volume Now Roughly 20% of Google's — mikeflache · 2026-11-10