Home / News

Claude Fable 5.2 gets cheaper, but small business bots should still default to smaller models

Claude Fable 5.2 costs $8 per million input tokens and $40 per million output tokens. That makes Anthropic's top model cheaper, but it is still too expensive to run every step of a small business AI assistant. For lead intake, invoice parsing or support replies, a smaller Claude model should do the routine work, and Fable should be called only for the rare hard case.

Anthropic released Fable 5.2 and Mythos 5.2. They are the same model, and Mythos is reserved for vetted cyber and biomedical teams. The model generates output up to 30% faster than 5.1. An optional fast mode at $16/$80 adds up to 2.5x speed. On AutomationBench, a business workflow test, Fable 5.2 scores 51.4% against 42.5% for Opus 5.5.

What this means for small business automation

Anthropic itself recommends starting with Opus 5.5 and stepping down to Sonnet or Haiku where possible. That matches how I build pipelines. Haiku sorts emails and pulls fields from invoices, and a heavier model handles contracts or long tender documents. Fable's new Advisor role fits this pattern: a cheap model consults Fable only when it gets stuck, so top rates apply to a handful of calls instead of thousands. The token bill is usually a minor line next to integrations and setup, which is what drives the cost of an automation project.

Why it matters: Estimate your monthly token volume per process before picking a model. For most small businesses the right setup is a cheap model by default and Fable as the escalation path.

Source: Hacker News