Wes Roth · 2026-09-03 · notable
Wes Roth — 'this JUST became the #1 AI model' on Claude Fable 5.1 effort
Wes Roth's September 3 episode builds four games with Claude Fable 5.1, including a social deduction game whose players are LLMs. His main claim: almost all of it ran on the model's low and medium effort settings.
Wes Roth builds four playable games with Claude Fable 5.1 and says the lower effort settings did almost all of the work.
What is it?
'this JUST became the #1 AI model...' went up on the Wes Roth channel on September 3, 2026. The episode is a build log rather than a benchmark run. Roth describes making Blood Grid, a voiced fantasy-football game with its own AI engine, a headless simulator and a Bradley-Terry balance bench; a bullet-time subway shooter; a self-playing isometric extraction game; and The Gilded Mansion, a social deduction game whose players are LLMs.
How does it work?
The claim the episode centres on is about effort settings. Roth says almost all four projects were built on low and medium effort, and suggests anyone who finds the model overrated has the slider set too high. Anthropic released Claude Fable 5.1 on September 1, 2026 with five effort levels — low, medium, high, x-high and max. Roth also says the model warned him about the knock-on effects of his own rule changes before they broke a game.
Why does it matter?
Effort level is the practical cost lever on Claude Fable 5.1, which Anthropic prices at $10 per million input tokens and $50 per million output tokens. If low effort really handles multi-file game builds, the same work costs far less than a max-effort run. This is one person's build log and his own opinion, not a measured comparison — treat the effort claim as something to test on your own workload.
Who is it for?
developers building with Claude Fable 5.1