claude-fable-5 API — per-token pricing and code examples
Price per 1M tokens
Cache rates apply when the provider reports cache tokens.
72% offAbout this model
claude-fable-5 is served through OpenAI Chat Completions, OpenAI Responses, Anthropic Messages. You pay $2.80 per 1M input tokens and $14.00 per 1M output tokens. That is 72% below the official price of $10.00 and $50.00. Cache reads cost $0.28 per 1M tokens when the provider reports cached tokens. It is included in budget packs from $99.00 up.
Where Claude Fable 5 fits
Claude Fable 5 is a Fable model from Anthropic, released on June 9, 2026 and now marked legacy. Anthropic lists its retirement as not sooner than June 9, 2027, and its training knowledge ends in January 2026. Fable 5 handles text and image input with text output, a 1M-token context window, and up to 128K output tokens per synchronous response.
Anthropic's legacy page for Fable 5 does not describe new use cases; it recommends moving to Claude Fable 5.1. Fable 5 remains useful for teams that validated their workflow on it and want to keep that baseline while they test Fable 5.1, and for agent code that still forces a particular tool, which Fable 5 accepts and Fable 5.1 does not.
Always-on thinking, one effort per request
Fable 5 runs adaptive thinking on every request, and there is no way to switch it off. Anthropic lists its thinking mode as "Adaptive (always on)": a request that sends thinking: {type: "disabled"} or between_tools comes back with a 400. Thinking text is hidden by default, because display defaults to omitted and the block arrives empty with only a signature.
Effort is the lever you control. The default is high, and Anthropic's guidance for Fable 5 reads: start with high for most tasks, use xhigh for the most capability-sensitive workloads, and step down to medium or low for routine work. Fable 5 does not support per-message effort. Anthropic states that models without per-message effort, including Claude Fable 5, return a 400 error when you try it, so set one output_config.effort value for the whole request.
On the tool side, Fable 5 is more permissive than its successor. Forced tool use is not on Anthropic's list of models that reject it, so tool_choice with a named tool still works. Structured outputs through output_config.format are supported too.
What it costs on Cent AI
Fable 5 bills against your prepaid USD balance per token, with the rates listed in this page's price box. Since thinking is always running, each request carries reasoning tokens, and Cent AI charges them at the output rate like the visible answer. Moving routine traffic from high to medium or low, as Anthropic itself suggests, is the main way to reduce that share.
On Fable 5 a prompt needs at least 512 tokens before it can be cached. You choose between Anthropic's five-minute default lifetime and an optional one-hour lifetime, and on Cent AI cache writes for both lifetimes are billed at the input rate. Only the cache counts that Fable 5 returns in its usage are billed at cache rates. Editing your tool list invalidates the whole cache, so keep tool definitions stable between calls.
Using it from your tools
Fable 5 can be called from /v1/messages, /v1/chat/completions and /v1/responses. Cent AI does not change your parameters silently. If your code sets temperature, top_p or top_k to anything other than the default, Fable 5 answers with 400, and from API version 2026-10-21 the field is refused at any value. A prefilled closing assistant message is also rejected.
A /v1/messages request with a forced tool, which Fable 5 still allows:
{
"model": "claude-fable-5",
"max_tokens": 4096,
"output_config": { "effort": "high" },
"tool_choice": { "type": "tool", "name": "save_summary" },
"tools": [{
"name": "save_summary",
"description": "Store a summary of the document",
"input_schema": {
"type": "object",
"properties": { "summary": { "type": "string" } },
"required": ["summary"]
}
}],
"messages": [{ "role": "user", "content": "Summarise the attached contract" }]
}
Specification
- Context window
- —
- Input and output
- Text
- Endpoints
- /v1/chat/completions, /v1/responses, /v1/messages
- Smallest pack
- $99.00
Latency and error rate per model are on the status page. status.cent-ai.dev
What a request costs
Open the full calculatorCost per request
Estimate
claude-fable-5An estimate at price version 5. Your bill comes from the usage the provider reports for each request.
Save $1.7143Code
curl https://api.cent-ai.dev/v1/chat/completions \
-H "Authorization: Bearer $CENT_AI_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "claude-fable-5", "messages": [{"role": "user", "content": "Hello"}]}'Set CENT_AI_KEY to a key from the dashboard. OpenAI SDKs use https://api.cent-ai.dev/v1; Anthropic SDKs use https://api.cent-ai.dev.
Price history
Version 5, in effect since Oct 31, 2026.
Effective
per 1M tokens
Inputold $2.80 new $2.80Cache readold $0.28 new $0.28Cache writeold $3.50 new $2.80Cache write 1hold $5.60 new $2.80Outputold $14.00 new $14.00Effective
per 1M tokens
Inputold $1.70 new $2.80Cache readold $0.17 new $0.28Cache writeold $2.125 new $3.50Cache write 1hold $3.40 new $5.60Outputold $8.50 new $14.00Effective
per 1M tokens
Inputold $5.75 new $1.70Cache readold $0.575 new $0.17Cache writeold $7.1875 new $2.125Cache write 1hold $11.50 new $3.40Outputold $28.75 new $8.50Effective
per 1M tokens
Inputold $1.90 new $5.75Cache readold $0.19 new $0.575Cache writeold $2.375 new $7.1875Cache write 1hold $3.80 new $11.50Outputold $9.50 new $28.75
Compare with
- deepseek-v4.1-flash-uncensoredInput$0.156
- claude-sonnet-5.5Input$0.58
- gpt-6-astraInput$3.20
- claude-opus-5Input$1.15
Questions
How much does claude-fable-5 cost on Cent AI?
$2.80 per 1M input tokens and $14.00 per 1M output tokens. Cache reads cost $0.28 and cache writes $2.80 per 1M tokens. Each request is rounded up once.
Is there a minimum charge?
A successful request costs at least $0.00002. A request that ends early, because the client left or the output was cut off, pays only for its tokens.
How do I pay?
Top up a prepaid USD balance with QRIS or USDT/USDC, or buy a budget pack once. Nothing renews automatically.
What if the provider does not report usage?
The request is billed from an estimate and marked estimated on the Requests page. An estimate never contains cache tokens.
Which endpoint do I call?
/v1/chat/completions, /v1/responses, /v1/messages. The OpenAI base URL is https://api.cent-ai.dev/v1.
Try claude-fable-5 with one key
One prepaid balance for every model. Pay with QRIS or USDT/USDC.