ANTHROPIC / MODEL PROFILE
ClaudeA workspace for demanding documents.
Fable 5.1 · Opus 5
Worth comparing for sustained research, document analysis, office deliverables, and large coding projects.
Research checked September 5, 2026
THE ASSESSMENT
Documented capabilities and limitations, with practical interpretation.
Tokens are small units used to represent inputs and outputs. API rates are usage charges for software, separate from chat subscriptions. Prices are in US dollars, as checked for this edition.
Read our research method →Where it stands out
Documents with visual structure
Anthropic describes Fable’s ability to interpret diagrams, charts, and tables within documents and PDFs. That makes it a useful candidate when a summary must account for exhibits and tables as well as prose.
Anthropic · Fable capabilities and safeguards ↗Long projects and large inputs
Fable 5.1 supports a one-million-token context window and up to 128,000 output tokens. Anthropic positions it for extended coding, multistep research, and document, spreadsheet, and slide work.
Anthropic · Fable 5.1 technical guide ↗A useful tier below the flagship
Anthropic recommends starting with Opus 5 for most workloads, escalating to Fable when demanding reasoning or extended tasks justify it. That is a useful reminder to test the everyday option first.
Anthropic · Fable 5.1 technical guide ↗Where to be careful
Retention needs a specific review
Fable requires 30-day data retention for safety monitoring by default. Anthropic describes exceptions for eligible enterprise customers. Do not carry over a zero-retention assumption from another Claude model or agreement.
Anthropic · Fable capabilities and safeguards ↗Some requests use a different model
Cybersecurity and biology safeguards can block or reroute requests. Fable and restricted Mythos share an underlying model, but their deployed capabilities differ because of these safeguards.
Anthropic · Mythos access and safeguards ↗More capacity is not automatically more value
Fable’s API rates are $10 input / $50 output per million tokens, versus Opus 5’s $5 / $25. Thinking is always on in Fable. Compare turnaround, accuracy, and revision effort on your own documents.
Anthropic · Release notes and pricing ↗Anthropic · Fable 5.1 technical guide ↗OUR SUGGESTED TRIAL
Try it on your kind of work.
Use a fictional agreement and two amendments. Request a table of obligations with exact supporting passages, unresolved conflicts, and questions for the reviewer. Then ask for a plain-English client explanation.
What to look for
Check whether an amendment superseded the clause the model quoted. Evaluate omitted qualifications and the quality of the rewrite separately.
An editorial test idea, not a reported benchmark result. Use public, fictional, or properly approved material.What else to compare
Opus 5 is the sensible baseline for this comparison. Sonnet 5 is another lower-cost tier; do not assume Fable’s specifications, retention terms, or safeguards apply to every Claude model.
Anthropic · Fable 5.1 technical guide ↗Anthropic · Release notes and pricing ↗No model in this guide replaces source checking, professional judgment, or a review of the application’s data terms.
Sources & further reading
Primary documentation checked for this edition. Provider claims are not an independent head-to-head test.