Why the Government Just Killed Claude Fable 5
TL;DR
A government directive appears to have killed Fable 5 almost immediately: Anthropic says it received a 5:21 Eastern legal order to suspend Fable 5 and Mythos 5 for foreign nationals, and the practical result was disabling the model for everyone.
Anthropic says it is complying, but does not agree: Riley highlights Anthropic's statement that the letter gave no specific technical details and that the action was not transparent, fair, or grounded in clear facts.
Fable 5 was the public, nerfed version of Mythos 5: Riley describes Mythos as a five-star frontier model and Fable 5 as a four-star release with safeguards that could reroute users if they pushed into advanced research territory.
The core allegation is a jailbreak, not a model failure disclosed in detail: According to Anthropic's understanding of the order, the government believes someone found a way to bypass Fable 5's safeguards, though no company or exploit was publicly named.
Users were already treating the model like magic: Riley cites examples from Todd Sanders, Peter Yang, Theo, and his own tests, including live feature-building from customer calls, realistic racing games, spatial planning apps, Minecraft clones, and rebuilding Lovable's mobile app.
The economics were already wild: Riley says API use could run $300 to $600 per hour, with his own Lovable clone costing roughly $200 to $300 for 9 to 10 prompts, which fuels speculation that Fable 5 was never sustainable as a broadly available product.
The Breakdown
Anthropic abruptly pulled Claude Fable 5 just 76 hours after launch, saying the US government ordered access suspended over a suspected jailbreak, and Riley Brown argues it may be the biggest AI access shock since Sam Altman's 2023 firing. He frames Fable 5 and Mythos 5 as the most fun and capable models people had touched yet, then shows why the shutdown also exposed a brutal reality: these systems can cost $300 to $600 an hour to run.
Was This Useful?
Share
Keep Reading
Make Alcreon Yours
Tune your feedFive quick questions, and the feed ranks what matters to you first.Or just get notified
The weekly Echo. Signal worth keeping in your inbox.
Every new piece, announced on X.
Read Next
See all
Playbook
Cheap Models, Hard Tasks
Most agent workflows route every step to the frontier model by default. The bill scales with how chatty the agent gets, even when most steps don't need that brain.

Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.

Playbook
The Art of Tasteful Prompting
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.