Anthropic · Claude Fable 5.1
claude-fable-5-1Available from the provider.
- Released
- Deprecatednot announced
- Retirednot announced
Summary
Claude Fable 5.1 is Anthropic's most expensive model, at $10 per million input tokens and $50 per million output tokens with a 1M-token context window and 128K max output. Anthropic aims it at the hardest reasoning and the longest-running agentic jobs, not at everyday work.
Fable sits above Opus in the Claude line. Anthropic's own guidance is to start on Claude Opus 5 for most workloads and to reach for Fable 5.1 when evaluations on Opus 5 at higher effort still fall short. Fable 5.1 was released on 1 September 2026 and succeeds Claude Fable 5 at the same input and output price, with a June 2026 knowledge cutoff and cache reads at a quarter of the Fable 5 rate.
It accepts text and image input and returns text. Adaptive thinking is always on and cannot be disabled; effort defaults to high, and Anthropic rates its latency as the slowest in the current line-up. Three changes break code written for Fable 5: forced tool use returns an error, earlier models cannot read its thinking blocks, and editing earlier turns invalidates those blocks.
No deprecation has been announced. Anthropic commits to keeping Claude Fable 5.1 available until at least 1 September 2027.
Tags
Editorial tags reflect our assessment; documented tags are backed by the sources below.
Specifications
- Input price
- $10.00 / M tok
- Output price
- $50.00 / M tok
- Price class
- Flagship
- Context window
- 1M tokens
- Max output
- 128,000 tokens
- Released
- 1 September 2026
- Input modalities
- text, image
- Output modalities
- text
- Knowledge cutoff
- June 2026
- Tool use
- Yes
- Structured output
- Yes
- License
- unknown
- Maker
- Anthropic
Use it for
- Run long-horizon agentic tasks that have to hold a plan across many tool calls without losing the thread.
- Take on reasoning work where Claude Opus 5 at high effort still fails your evaluations.
- Process very large inputs in one pass, using the 1M-token context window and cache reads at $0.25 per million tokens.
- Move off Claude Fable 5, which costs the same per token but has an older knowledge cutoff and four times the cache read price.
Avoid it when
- Cost per call is a constraint; at $10 / $50 this is five times the input price of Claude Sonnet 5 and ten times that of Claude Haiku 4.5.
- You need a fast response per turn; thinking is always on and Anthropic rates its latency as the slowest of the current models.
- Your integration relies on forced tool use, which returns an error on this model.
- Your input includes audio or video; Claude Fable 5.1 accepts text and images only.
Alternatives
What to use instead. Curated suggestions are written by our editors for this model; derived ones are ranked automatically from tags, tier and price class.
- Claude Opus 5curated
Anthropic
claude-opus-5Half the price at $5 / $25 with the same 1M-token window, and the model Anthropic names as the starting point for most workloads.
In $5.00 / M tokOut $25.00 / M tok
ActiveTextVisionChat+7 - Claude Fable 5curated
Anthropic
claude-fable-5The previous Fable release at the same $10 / $50, worth keeping only if you depend on behaviour that changed in 5.1.
In $10.00 / M tokOut $50.00 / M tok
ActiveTextVisionChat
Change history
Nothing has changed since this model was first recorded — no status change and no price change.
Price history
No price changes recorded yet.
Sources & verification
Facts
Status, dates, pricing and specifications were verified against these pages.
Last verified: (2026-09-23 04:30) · ✓ verified
Editorial profile
The summary, the advice and the curated alternatives are our assessment, not a provider claim.
Editorial profile reviewed on by llm-researcher.