ANTHROPIC MODEL GUIDE
Claude Fable 5.1 API
Anthropic's model for demanding reasoning, long-running coding agents and multi-step research, with image understanding and always-on adaptive thinking.
Choosing and integrating
Claude Fable 5.1 API pricing breakdown
Claude Fable 5.1 (claude-fable-5-1) is billed by token usage. Prices below are USD per 1M tokens. LLMFly AI rates already include the 0.5x group multiplier; do not apply it again. Provider reference prices are separate. These rates come from the site's Model Plaza snapshot; confirm live catalog and account rates before calling.
Example: a request with 10,000 uncached input tokens and 1,000 output tokens costs approximately $0.075 in LLMFly AI token charges. This excludes cache writes, tools, retries, and other additional fees.
- Platform and provider prices use their own input-token tiers. Do not apply provider long-context multipliers to platform prices.
- Not listed does not mean free. Account for cache reads, cache writes, and uncached input in their respective billing categories.
- Token cost = tokens in each billing category × its price per million ÷ 1,000,000. Add tools and other charges separately.
| Price source | Input tokens per request | Input | Output | Cache write | Cache read | 1h cache write |
|---|---|---|---|---|---|---|
| LLMFly AI | Single rate | $5 | $25 | $6.25 | $0.125 | $10 |
| Provider reference | Single rate | $10 | $50 | $12.5 | $0.25 | $20 |
Pricing tiers and additional charges
Model Plaza now lists 1-hour cache writes at $20 per million tokens officially and $10 on the platform after the 0.5x multiplier. Confirm cache hits, write duration, and tools against actual usage.
Claude Fable 5.1 capabilities and workloads
Anthropic's model for demanding reasoning, long-running coding agents and multi-step research, with image understanding and always-on adaptive thinking.
Difficult coding and research where a slower, more expensive model is justified by task results.
- Tool use and structured outputs; use automatic rather than forced tool selection.
- Native beta features include per-message effort and readable progress updates between tool calls.
Limits to check before migration
- Forced tool_choice any/tool returns HTTP 400. Manual thinking budgets and disabling thinking are unsupported.
- Assistant prefilling and non-default temperature, top_p or top_k return HTTP 400 in the native API.
- Earlier models cannot read its thinking blocks. Editing earlier conversation content can invalidate preserved thinking; consult the migration guide.
Safety fallback and data retention
Anthropic applies additional safeguards to some cybersecurity and biology requests, including refusals or fallback; API clients must handle fallback settings. The vendor requires 30-day safety-monitoring retention by default, with exceptions for eligible enterprises. This does not establish zero retention or automatic fallback on an LLMFly route; confirm the actual policy before integration.
Comparing with Claude Fable 5
At this review, Fable 5.1 and Fable 5 have the same base input, output, and cache-write rates. Base cache reads are $0.25 versus $1 per million tokens. With the same multiplier, 5.1 cache reads cost 75% less; that is not a 75% saving on the whole request.
Use the exact model ID when connecting
Set the model field to claude-fable-5-1; do not infer its spelling from the display name. Configure the API address and key using the access documentation; the snippet only shows the model field, not a complete request.
- Send a minimal text request and inspect the response and usage.
- Test streaming, tool calls, and structured output separately if your application needs them.
- Check retry behavior and spending limits before increasing traffic.
{
"model": "claude-fable-5-1"
}Vendor capability versus gateway access
Claude Fable 5.1's specifications above were checked in vendor documentation. No paid inference requests or account-route modality, tool, or beta-parameter tests were run in this review. Validate your application before launch; documentation review is not end-to-end compatibility certification.
Frequently asked questions
What should I put in the model field?
Use the exact ID claude-fable-5-1, not the display name.
How is the discount calculated?
The current 0.5x multiplier is applied to the base rate, giving 50% off the catalog's official reference. Amounts are shown per million tokens.
Do these prices include every possible fee?
Not necessarily. Check cache-write duration, tools, and other route charges separately. Use the full pricing table and actual usage records to reconcile a bill.
What are the context and output limits?
Vendor documentation lists 1M context and 128K maximum output; confirm gateway route limits separately.
Use the Claude Fable 5.1 API
LLMFly AI provides this model. Use the exact ID claude-fable-5-1 and confirm current account pricing in Model Plaza.