Released June 9, 2026, Claude Fable 5 is the first publicly available Mythos-class model from Anthropic. Here is what it can do, what it costs, and what your team needs to do right now.
Anthropic spent months keeping its most capable model locked behind closed doors. On June 9, 2026, that changed. Claude Fable 5 is the first publicly available model in the Claude 5 family and the first to carry what Anthropic calls a Mythos-class capability tier, a level that sits above Opus, which had been the top of the line since 2024. This is not a decimal-point refresh. For business teams watching the AI model race from the sidelines, this is the update worth paying attention to.
What Exactly Is the Mythos Tier?
Fable 5 arrives eight days after Anthropic confidentially filed for an IPO, and two months after the company chose not to release this architecture publicly, routing it instead into Project Glasswing, a defensive security program whose partners include Microsoft, Google, and NATO-affiliated agencies.
Claude Fable 5 and Claude Mythos 5 share the same underlying model. Fable 5 is the version made safe for general use. Mythos 5 has its safeguards lifted in specific areas and ships first through the restricted Project Glasswing program.
In plain English: the same engine powers both. Fable 5 is the version your team can sign into today. Mythos 5 is the unlocked version available only to vetted infrastructure providers and security researchers.
The Claude family covers three named size tiers, Opus, Sonnet, and Haiku, introduced with Claude 3 in March 2024, now joined by the Mythos-class tier sitting above Opus.
What It Can Do That Older Models Cannot
The headline capability is long-horizon work: tasks that run for hours and span many steps without a human nudging the model back on course.
The benchmark numbers back that up. Fable 5 scored 80.3% on SWE-Bench Pro, against 69.2% for Opus 4.8. That gap is significant for any business using AI for software development.
Real-world results are even sharper. Stripe reported that Fable 5 compressed months of engineering work into days, completing a large Ruby codebase migration that would have taken a team over two months. On the analytical side, it is the first model to clear 90% on Hex's analytical benchmark, built from long multi-stage data tasks, and it posted the highest score of any model on Hebbia's finance benchmark.
For research-heavy industries, Anthropic also reports the model accelerated parts of drug design work by around ten times.
Fable 5 also landed in GitHub Copilot on day one, making it the first Mythos-class model available directly inside a developer workflow that millions of teams already use.
The Safety Architecture: How Anthropic Made This Releasable
This matters most for businesses thinking about governance and risk. The model ships just days after Anthropic publicly warned that frontier AI is becoming dangerously capable, and it carries the most aggressive safety scaffolding the company has put on a general release.
Fable 5 routes every conversation through classifiers, separate AI systems that watch for potential misuse. When a query touches offensive cybersecurity, certain biology and chemistry territory, or attempts to distill the model, Fable 5 does not answer it. Claude Opus 4.8 answers instead. Anthropic says more than 95% of Fable sessions involve no fallback at all, so for most work you will never notice the seam.
One data retention point your legal or compliance team should note: Anthropic designates Fable 5 and Mythos 5 as covered models with 30-day data retention. GitHub's Copilot changelog confirms that Fable 5 prompts and outputs may be retained for up to 30 days to operate Anthropic's safety classifiers. Other Claude models in Copilot, including Opus 4.8, continue under zero data retention there.
If your workflows involve confidential data, decide which model tier is appropriate before your developers start defaulting to Fable 5.
Pricing and Availability
The practical facts for developers building on the Claude API:
- Model ID:
claude-fable-5 - Pricing: $10 per million input tokens, $50 per million output tokens
- Context window: 1 million tokens
- Max output: 128,000 tokens
That pricing is exactly double Opus 4.8. The model is available immediately through the Claude API and Amazon Bedrock, and is included free for Claude Pro, Max, Team, and Enterprise users during an introductory window from June 9 through June 22, 2026. After June 23, usage credits apply.
Fable 5 is available across the Claude API, AWS Claude Platform, Amazon Bedrock, Vertex AI, Microsoft Foundry, and GitHub Copilot. That wide distribution on day one means most teams already using cloud AI services can test it immediately with no new vendor agreements required.
What This Means for Business Teams Right Now
The competitive picture has shifted. In blind evaluations run by Surge, Microsoft's independent human rating partner, Microsoft's MAI-Thinking-1 was preferred over Anthropic's Claude Sonnet 4.6, the first credible third-party evaluation showing a Microsoft-built model competitive with Anthropic's frontier. Fable 5 is Anthropic's strongest answer yet.
Here are the practical steps for any business looking at this release:
- Test it before June 22. If your team is on a paid Claude plan, Fable 5 is included free until June 22. Run your hardest, most time-consuming workflows through it now and measure the difference against Opus 4.8 before committing budget.
- Audit your data policies before deploying. The 30-day retention policy on Mythos-class models is a meaningful change from Opus 4.8's zero-retention profile on some platforms. Know what data your prompts contain before they hit the new model.
- Focus on long-horizon tasks first. Fable 5's biggest gains are in multi-step, sustained work: codebase migrations, complex data analysis, lengthy research synthesis. Using it as a drop-in replacement for quick chat queries is not where the return is.
- Plan your API integration carefully. The model ID is
claude-fable-5, and sensitive queries fall back silently to Opus 4.8. If your application depends on consistent capability levels across all queries, test edge-case inputs early to understand where the fallback triggers.
Anthropic has moved the ceiling, made that new ceiling accessible on platforms enterprises already use, and been unusually transparent about the tradeoffs. The window to evaluate is open right now.
How 247techify can help
At 247techify, we help businesses evaluate new AI models, integrate them into real workflows, and build the governance policies that keep sensitive data protected. If you want to know whether Claude Fable 5 belongs in your stack and how to deploy it safely, get in touch at https://www.247techify.com/ and let's talk it through.
