Claude Fable 5 Explained
A plain-language guide to Anthropic's newest public model: what it does, what it costs, the compliance rules, and how it compares.
Status: Claude Fable 5 is generally available worldwide as of July 1, 2026, on the Claude Platform, Claude.ai, Claude Code, and Cowork (Anthropic). This guide explains Claude Fable 5 in plain terms. It covers the model's features, price, availability, safety design, compliance rules, and how it stacks up against other models.
Anthropic released Claude Fable 5 on June 9, 2026. It is the firm's most capable public model for software engineering, knowledge work, vision, and scientific research. It also works on its own for longer than earlier Claude models.
Fable 5 sits at the top of Anthropic's public lineup. Below we break down each part so you can decide if it fits your work. We use Anthropic's own sources for every model claim.
Reviewed by Jonathan West, Founder of Layer3 Labs, on July 1, 2026. We research using primary vendor and regulator sources.
What Is Claude Fable 5?
Claude Fable 5 is Anthropic's most capable public model for software engineering, knowledge work, vision, and scientific research (Anthropic). It reads long documents, writes code and text, and understands images.
Anthropic released it on June 9, 2026. It replaces earlier Claude models as the top public option for hard tasks.
One trait sets it apart. Fable 5 works on its own for longer than previous Claude models. This helps with multi-step jobs like reviewing a full contract set or debugging a codebase.
- Software engineering: writes, reviews, and fixes code across many files.
- Knowledge work: drafts, summarizes, and analyzes long documents.
- Vision: reads scanned files, diagrams, and images.
- Scientific research: supports analysis and hypothesis work.
Deciding whether Claude Fable 5 fits your workflows and compliance needs? We can map the model to your team and flag the 30-day retention tradeoff.
Book a ConsultationKey Features and Capabilities
Claude Fable 5's main strength is autonomous work over longer tasks (Anthropic). It keeps going through multi-step jobs with less hand-holding than earlier models.
The model spans four core areas: coding, knowledge work, vision, and science. That range lets one model handle drafting, review, and analysis in the same workflow.
It ships across Anthropic's full surface. You can use it in the Claude Platform, Claude.ai, Claude Code, and Cowork (Anthropic).
- Longer autonomous runs than previous Claude models.
- Strong on complex, multi-file software engineering.
- Handles long documents in one pass.
- Reads images and scanned pages (vision).
- Available in Claude Code for developer workflows.
Claude Fable 5 Context Window Size
Claude Fable 5 uses a large context window, and a 1M-token context option is available for long-document work (Anthropic). The context window is how much text the model can hold in mind at once, counting both your input and its output.
A bigger context window matters when you feed the model a lot at once. Think a full contract set, a long codebase, or months of chat logs. With a 1M-token window, Fable 5 can reason across all of it in one pass instead of losing the earlier parts.
The 1M-context option costs more and is not needed for short tasks. Use the standard window for everyday prompts. Reach for the 1M-token setting only when the job truly spans a large body of text, since you pay for every token in the window.
- Context window: the total text the model holds at once (input plus output).
- A 1M-token option supports very long documents and codebases.
- Best for whole-contract review, large codebases, and long histories.
- Costs more per run: use it only when the task really needs the length.
- For short prompts, the standard window is cheaper and just as accurate.
Pricing and Access
Claude Fable 5 costs $10 per million input tokens and $50 per million output tokens (Anthropic). Input is what you send in. Output is what the model writes back.
That means a long brief or a full code review costs cents, not dollars, per run. Output tokens cost five times more than input, so concise prompts and outputs save money.
You reach the model through Anthropic's own products today. Cloud marketplaces are re-enabling access in stages, covered further below.
- Input: $10 per million tokens.
- Output: $50 per million tokens.
- Access now: Claude Platform, Claude.ai, Claude Code, Cowork.
- Cloud: AWS Bedrock, Google Cloud, and Microsoft Foundry re-enabling in stages.
Best Uses for Business
Claude Fable 5 fits document-heavy and code-heavy roles best. Its knowledge-work and vision strengths help teams that read, draft, and review at volume.
The model suits regulated fields too, with one caution on data retention we cover below. Match the use case to the role for the clearest value.
- Software teams: multi-file coding, review, and debugging in Claude Code.
- Legal: contract review, memo drafting, and discovery summaries.
- Finance and accounting: statement analysis and document extraction.
- Medical and dental practices: notes and intake summaries under a HIPAA BAA.
- Real estate and property management: lease review and listing drafts.
- Insurance: claim file review and policy summaries.
Limitations and Risk Factors
Claude Fable 5's main limits are its safety fallback and a mandatory data retention rule (Anthropic). Both shape how and where you can use it.
Safeguards block high-risk cyber and bio requests. When that happens, the model falls back to Claude Opus 4.8. Anthropic reports more than 95% of Fable 5 sessions never trigger a fallback.
The retention rule matters most for regulated work. As a Mythos-class model, Fable 5 carries a mandatory 30-day safety retention on business accounts. Full zero-data-retention is not available on it.
- Safeguards block high-risk cyber and bio requests.
- Blocked requests fall back to Claude Opus 4.8.
- Fewer than 5% of sessions trigger a fallback (Anthropic).
- No full zero-data-retention option today.
- Human review is still required for every output.
Compliance and Data Handling
Claude Fable 5 inherits Anthropic's SOC 2, ISO 27001, and ISO 42001 certifications, with a HIPAA BAA on the API and Enterprise plans (Anthropic). These cover the platform, not your own legal duties.
One rule is easy to miss. As a Mythos-class model, Fable 5 carries a mandatory 30-day safety retention on business accounts. Anthropic uses that data only for safety, then deletes it.
This means full zero-data-retention is not available on this model. If a contract or regulator demands zero retention, Fable 5 does not meet it today. Confirm your agreements allow a 30-day safety window first.
- Certifications: SOC 2, ISO 27001, ISO 42001.
- HIPAA BAA on the API and Enterprise plans.
- Mandatory 30-day safety retention on business accounts.
- Data used only for safety, then deleted.
- Full zero-data-retention is not available.
How Claude Fable 5 Compares
Claude Fable 5 is Anthropic's top public model, with a safety fallback to Opus 4.8 and no zero-retention option (Anthropic). Its main peers are Opus 4.8, GPT-5.6, and Claude Mythos 5.
Opus 4.8 is the safer fallback model that Fable 5 hands off to on blocked requests. GPT-5.6 is OpenAI's newest family, still in a limited preview as of July 1, 2026. Mythos 5 is the same base model as Fable 5 but with safeguards lifted for vetted partners.
Use the table to match a model to your need. For most businesses, Fable 5 is the deployable choice because it is generally available worldwide today.
- vs Opus 4.8: Opus 4.8 is the fallback for blocked requests; Fable 5 is the higher-capability default.
- vs GPT-5.6: GPT-5.6 is a limited preview open only to vetted organizations; Fable 5 is GA worldwide.
- vs Mythos 5: Mythos 5 is the restricted, safeguards-lifted version for vetted partners; Fable 5 is the public option.
How to Access Claude Fable 5
Claude Fable 5 is generally available worldwide as of July 1, 2026, through Anthropic's own products (Anthropic). You can use it in the Claude Platform, Claude.ai, Claude Code, and Cowork.
Availability had a brief pause. A June 12 export-control suspension stopped some access, then the US Commerce Department lifted it, restoring GA worldwide on July 1, 2026.
Cloud marketplaces are coming back in stages. AWS Bedrock, Google Cloud, and Microsoft Foundry are re-enabling Fable 5 access over time. Check each marketplace for current status.
- Anthropic products: Claude Platform, Claude.ai, Claude Code, Cowork.
- API: send data under a HIPAA BAA where required.
- Cloud: AWS Bedrock, Google Cloud, Microsoft Foundry re-enabling in stages.
- June episode: export-control suspension lifted, GA restored July 1, 2026.
Frequently Asked Questions
- Claude Fable 5 is Anthropic's most capable public model for software engineering, knowledge work, vision, and scientific research. Anthropic released it on June 9, 2026, and it works autonomously for longer than earlier Claude models.
- Claude Fable 5 costs $10 per million input tokens and $50 per million output tokens (Anthropic). Input is what you send; output is what the model writes back. Output costs five times more, so concise prompts save money.
- Yes. Claude Fable 5 is generally available worldwide as of July 1, 2026, on the Claude Platform, Claude.ai, Claude Code, and Cowork. A June 12 export-control suspension was lifted by the US Commerce Department before GA.
- Claude Fable 5 uses a large context window, with a 1M-token context option available for long-document and large-codebase work (Anthropic). The context window is the total text the model holds at once, counting both input and output. Use the 1M-token option only when a task truly spans a large body of text, since you pay for every token in the window.
- A June 12, 2026 export-control suspension briefly limited access to Claude Fable 5. The US Commerce Department then lifted it, and Anthropic restored GA worldwide on July 1, 2026. Cloud marketplaces are re-enabling in stages.
- Safeguards in Claude Fable 5 block high-risk cyber and bio requests and fall back to Claude Opus 4.8. Anthropic reports more than 95% of Fable 5 sessions never trigger a fallback.
- Claude Fable 5 offers a HIPAA BAA on the API and Enterprise plans, and inherits SOC 2, ISO 27001, and ISO 42001 (Anthropic). The certifications cover the platform, not your own compliance duties, which still apply.
- No. As a Mythos-class model, Claude Fable 5 carries a mandatory 30-day safety retention on business accounts, so full zero-data-retention is not available. Anthropic uses the data only for safety, then deletes it.
- Claude Fable 5 is the higher-capability default, and Claude Opus 4.8 is the safer model it falls back to when a request is blocked. Both run in Anthropic's products.
- Claude Fable 5 is generally available worldwide, while GPT-5.6 is in a limited preview open only to a small set of vetted organizations as of July 1, 2026. For most businesses, Fable 5 is the deployable choice today.
- Claude Fable 5 and Claude Mythos 5 share the same base model, but Fable 5 keeps safeguards on while Mythos 5 lifts them for vetted partners in Project Glasswing. Fable 5 is the public option.
Put Claude Fable 5 to Work in Your Business
Not sure how Claude Fable 5 fits your workflows or compliance rules? Book a free 30-minute Layer3 Labs AI workflow audit and we will map it to your team.
Book Your Free AI Workflow Audit