About this model
Claude Fable 5.2 is built for demanding reasoning and long-horizon agentic work: agent sessions that run for hours, multistep deep research, and analysis carried through to a finished document, spreadsheet, or deck. It takes text, image, and file inputs with text output, thinks adaptively (always on), and has a 1M-token context window with 128K output. Use it when evals on Claude Opus 5.5 at higher effort still fall short.
Pricing
| Rate | RODI | USD (ref.) | Unit |
|---|---|---|---|
| Text tokens | |||
| In | 6155.0 | ~ 8.00 | per 1M tokens (RODI/M, USD/M) |
| Out | 30774.8 | ~ 40.00 | per 1M tokens (RODI/M, USD/M) |
| Prompt caching | |||
| Cached | 153.9 | ~ 0.200 | per 1M tokens (RODI/M, USD/M) |
| Cache write (5 min) | 7693.7 | ~ 10.00 | per 1M tokens (RODI/M, USD/M) |
| Cache write (1 h) | 12310.0 | ~ 16.00 | per 1M tokens (RODI/M, USD/M) |
RODI prices include RodiumAi markup and upstream fees. USD figures are wholesale reference rates.
Providers & pricing
| Provider | Input | Output | Cached | Free tier | Status |
|---|---|---|---|---|---|
Anthropic Default All rates (5)
| 6,155 | 30,775 | 153.9 | - | No data |
AWS Priority 2 All rates (5)
| 6,155 | 30,775 | 153.9 | - | No data |
Runware Priority 3 All rates (5)
| 6,155 | 30,775 | 153.9 | - | No data |
Capabilities
API usage
Native Anthropic Messages API: use client.messages() or the anthropic package.
Examples
…Example response
…Tool use
…Same model through the OpenAI format
…RodiumAi SDK
Use messages() for Anthropic-native payloads; OpenAI chat mode remains available below.
…Use cases
- Long document work: summarise, compare and answer questions over contracts, reports and transcripts.
- Coding assistants and agents (RodiumAi Code, Claude Code, Cline, Continue) pointed at RodiumAi.
- Careful writing and editing in French and English.
- Tool-using agents: Claude calls your functions and chains the results.
- Image understanding: send
imagecontent blocks next to your text.
Notes
- Native endpoint:
POST https://api.rodiumai.io/v1/messages, the Anthropic Messages format. Authenticate withx-api-key: rd_sk_YOUR_KEY(aAuthorization: Bearerheader also works). - With the Anthropic SDKs set the base URL to
https://api.rodiumai.io: the SDK appends/v1/messagesitself. max_tokensis required by the Messages API.- Claude 4.7 and later models do not accept
temperature,top_portop_k. RodiumAi removes them from the request, so leave them out for these models. - The same model also answers on
POST https://api.rodiumai.io/v1/chat/completions(OpenAI format), shown above. Use whichever format your tool already speaks. - Billing is per token. The response
usageshows input and output tokens, and prompt caching is billed at the cached rate when your request uses it. - Streaming is available with
"stream": truein both formats.