Give Claude Fable 5.2 the work that takes more than a quick answer: investigate a problem, connect the evidence, and develop a result you can review.
Start with the prompt in the composer or choose an example.
Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.
30% off official pricing· Input / output tokens
One 1K-token prompt with a 1K-token answer at the cheapest tier.
Claude Fable 5.2 is Anthropic's model for demanding reasoning and long-horizon agentic work, released on November 9, 2026. Claude Fable 5.2 accepts text and images, with a one-million-token context window and up to 128,000 output tokens. On reAPI, start with native Messages or the OpenAI-compatible endpoint, and adjust effort to suit the task.
Real-world workflows and production use cases you can build and ship with this model.
Use Claude Fable 5.2 in an agent that investigates a failing service or develops a feature against acceptance criteria. Supply the relevant files and tools, and state what a finished result must include. Your application runs the tools; Claude Fable 5.2 supplies the reasoning and requests.
Build your first request
Give Claude Fable 5.2 the reports, screenshots and notes behind a difficult decision. Ask for conflicting claims and references to the supplied material, so you can check its conclusions. The large context window gives Claude Fable 5.2 room to compare that evidence.
Ask Claude Fable 5.2 to develop a decision memo, structured report or presentation content from your analysis. Specify the audience, evidence and format. Document tools belong in your application: Claude Fable 5.2 returns text and tool calls, rather than automatically creating office files.
Use Claude Fable 5.2 through Messages for native tool blocks, thinking signatures and streaming events. OpenAI clients can use the compatibility endpoint, with field mappings in the docs.
The Claude Fable 5.2 table separates input, output and caching. Compare actual usage on your tasks: a lower unit rate alone does not predict a lower session bill.
Use your reAPI key and balance for Claude Fable 5.2 and other supported models. Compare alternatives on your application's acceptance criteria.
Sign in to reAPI and open API keys. Store the key on your server, then check your available balance and the current Claude Fable 5.2 rates before sending a request.
OpenFor native Claude Fable 5.2 requests, use Messages with model claude-fable-5-2, messages and max_tokens. Existing OpenAI clients can use Chat Completions; the docs explain its field mappings.
OpenStart Claude Fable 5.2 with a task you can evaluate, stream the response, and inspect its stop reason. Adjust effort and the output budget using the result, token usage and latency you observe.
OpenClaude Fable 5.2 supports a 1M-token context window by default. Include the material relevant to your task and leave room for the response; more context still means more input to process.
Claude Fable 5.2 supports up to 128K output tokens. Thinking shares that allowance with visible text, so an overly small budget can end a response before an answer appears.
Choose low, medium, high, xhigh or max for Claude Fable 5.2. The default is high, and adaptive thinking stays on; use effort to tune how much work a task receives.
Claude Fable 5.2 can reason over supplied images alongside text. On reAPI, provide public HTTP(S) media URLs for screenshots and visual references; inline base64 media is not accepted.
Claude Fable 5.2 can request tools and continue from their results. Use automatic tool choice or disable tools for a turn; the official model does not support forced tool selection.
Claude Fable 5.2 supports prompt caching with separate read and write rates. Keep stable instructions and reference material consistent, and inspect usage counters to see whether a request actually used the cache.
Common questions about this model.
Compare Claude Fable 5.2 with the version your application already uses. The main migration questions are conversation state and tool compatibility; measure quality and total cost on your own tasks before moving traffic.
Comparison follows Anthropic's migration guide, reviewed November 10, 2026. Platform-specific behavior is not a promise that every optional tool has been verified on reAPI.
Explore more models in the same category.
Anthropic
From $2.80 per 1M tokens
Anthropic
From $2.80 per 1M tokens
Anthropic
From $7.00 per 1M tokens
Anthropic
From $7.00 per 1M tokens
curl --no-buffer --request POST \
--url https://reapi.ai/api/v1/messages \
--header "x-api-key: YOUR_API_KEY" \
--header "Content-Type: application/json" \
--data '{
"model": "claude-fable-5-2",
"messages": [
{
"role": "user",
"content": "Review this migration plan. Identify assumptions, propose a minimal verification checklist, and separate confirmed facts from open questions."
}
],
"max_tokens": 4096,
"output_config": {
"effort": "high"
},
"stream": true
}'Rejected synchronously with the failing field named. Check enum values, ranges, and URL fields — nothing is charged.
401 means a missing or invalid Bearer key; 402 means the reserve exceeds your balance. Manage keys and credits in the console.
Reference material and outputs pass automated moderation. Rejected tasks fail with a clear error and a full refund.
A task that reaches a failed state is never charged — the reserve refunds automatically. Keep the task id and retry when ready.
Try it in the playground or grab an API key to integrate now.