Claude Fable 5.2 now available on AI Gateway

Vercel added Claude Fable 5.2 to AI Gateway. Use anthropic/claude-fable-5.2 as the model name through the AI SDK, the Chat Completions, Responses, and Anthropic Messages APIs, or a coding agent connected to the gateway. To point Claude Code, Codex, Cursor, and similar tools at it, install the latest Vercel CLI and run npx vercel ai-gateway setup.

Anthropic describes Fable 5.2 as built for open-ended technical work, with gains over earlier Fable models in coding, reasoning, and vision. It can run long, ambiguous tasks unattended, act as an orchestrator that hands simpler tasks to smaller models, or serve as an advisor other Claude models consult on harder problems. Effort runs from low to max, and high is the API default.

Two conditions to know about. Fable 5.2 needs data retention: Anthropic doesn't support Zero Data Retention for it, so prompts and completions are kept for 30 days (and aren't used to train Claude). If your team has ZDR turned on, AI Gateway returns an error instead of sending the request. The model also has cybersecurity and biology safety classifiers enabled.

To keep requests served when the cybersecurity classifier declines one, add a models array under providerOptions.gateway. The gateway tries each model in order. In Vercel's example, a security-review prompt to Fable 5.2 falls back to Opus 5.5, then Sonnet 5.5. The option works the same way on every AI Gateway API format.