Claude Fable 5.2 Is Now Live in Filot
Anthropic's Claude Fable 5.2 is now live in Filot. On complex, long-running modeling tasks, it builds models nearly twice as fast as Opus 5.5.
Anthropic's Claude Fable 5.2 is now live in Filot. Workspaces with Fable 5.1 enabled move to Fable 5.2 automatically. For everyone else it remains opt-in, and nothing changes.
Compared with Opus 5.5, we see a few clear benefits. On complex, long-running modeling tasks, it builds models nearly twice as fast. Part of the reason is that it takes 40% fewer steps to reach the same result.
The gains are biggest when building from scratch, with the largest improvements on LBOs and comps. These are the builds where the model has to set up the structure itself, and where every early decision carries through to the output.
On its own, the model is fast. Combined with Filot's reviewer agent, we find it leads to significantly higher quality output.
For analysts, that means stronger adherence to the custom workflows you build in Filot, with a much quicker turnaround. That makes it a natural fit for Autopilot, our new product built to help analysts scale coverage quickly while following their own methodology.
Fable 5.2 carries the same data handling as the previous Fable model. To enable it for your organization, reach out to your Filot contact.
More from Filot
See all →Claude Opus 5.5 Is Now Available on Filot
Anthropic's Claude Opus 5.5 is now available on Filot. It builds more thorough financial models than Opus 5, in less time and at lower cost.
Fable 5.1 Is Now Available on Filot
Anthropic's newest frontier model is now available on Filot. It leads our internal evaluations on long-running financial modeling tasks, and discounted prompt cache reads make it cheaper to run than Fable 5.
Verifying AI Output Shouldn't Take Longer Than the Work Itself
AI only saves time if you can verify its work efficiently. Filot audits at two levels, deterministic validation checks and granular cell-by-cell traceability, so analysts review AI-generated models at the speed of reading, not the speed of searching.