Claude Fable 5: Anthropic Brings Mythos-Class Intelligence to the Public
Published: June 9, 2026
On June 9, 2026, Anthropic took a significant leap forward in the AI landscape by launching Claude Fable 5 — the first publicly available "Mythos-class" model. Two months after the private rollout of Mythos shook Wall Street and sparked intense debate about AI safety, Fable 5 delivers that same raw intelligence to enterprise customers and subscribers, wrapped in robust safety guardrails.
This is what you need to know about the model, what it can do, and where the boundaries are drawn.
What Is Claude Fable 5?
Claude Fable 5 is a Mythos-class model — the same underlying neural architecture as the highly restricted Claude Mythos that Anthropic debuted in April 2026. The difference? Fable 5 has safety filters applied on top, restricting the model's capabilities in high-risk domains like cybersecurity and biology.
Think of it as Mythos with training wheels — same engine, different rules of the road.
Anthropic describes it as "a Mythos-class model made safe for general use." For organizations that need the full uncapped power, Anthropic is simultaneously launching Claude Mythos 5, available initially through Project Glasswing and a trusted-access program.
Capabilities: What Fable 5 Can Do
Fable 5 is state-of-the-art across nearly every benchmark tested. Its strengths cluster around four key areas:
Software Engineering
This is where Fable 5 truly shines. Stripe reported that Fable 5 compressed months of engineering into days — performing a codebase-wide migration in a 50-million-line Ruby repository that would have taken a full team over two months by hand.
Key capabilities:
- Writes its own tests to validate work
- Handles multi-day autonomous coding sessions
- Implements designs with high fidelity
- Uses vision to compare outputs against designs
- Top performer on CursorBench, FrontierCode, and SWE-bench
Cursor called it "state of the art on CursorBench" — opening up "a class of long-horizon problems that were out of reach for earlier models."
Knowledge Work & Complex Analysis
For complex, multi-stage analytical tasks, Fable 5 sets new records. Analytics company Hex reported Fable 5 was the first model to break 90% on their core analytics benchmark — a 10-point jump over Opus 4.8.
Notable performance areas:
- Document-based reasoning and chart/table interpretation
- Senior-level financial analysis (top score on Hebbia's Finance Benchmark)
- Trading analysis across factual lookup, conceptual reasoning, and root-cause analysis
- Complex, long-running analytical tasks with strong judgment and attention to nuance
Rakuten noted: "At the highest effort, Fable reflects on and validates its own work. For us, that's what makes highly autonomous operations possible — the extra thinking pays for itself."
Vision
Fable 5 can extract precise numbers from detailed scientific figures and perform complex vision-based tasks like rebuilding a web app's source code from screenshots alone.
In a particularly striking demonstration, Fable 5 completed the game Pokémon FireRed using only raw game screenshots — no maps, navigation aids, or extra game-state information. Earlier Claude models needed a complex helper harness to manage the same task.
Memory & Long Context
Fable 5 stays focused across millions of tokens in long-running tasks and improves its outputs using its own notes. In testing with the deck-building game Slay the Spire, giving Fable access to persistent file-based memory improved its performance three times more than it did for Opus 4.8.
The Safety Story: Guardrails and Fallbacks
Fable 5 is the same base model as Mythos 5, but with significant safety measures:
- Domain restrictions: Queries flagged in cybersecurity, biology, chemistry, and distillation are automatically routed to Claude Opus 4.8 (a less capable model) instead
- No extra cost for fallbacks: You won't be charged Fable prices for routed-down requests
- Rare triggers: Early data shows at least 95% of Fable sessions run entirely on the model's own responses
- Red-teaming: Anthropic ran an external bug bounty that produced no universal jailbreaks in over 1,000 hours of testing. External red-teaming organizations also failed to find universal jailbreaks
- 30-day data retention: All traffic is retained for 30 days for safety monitoring — even for enterprises that previously had zero-retention agreements. Anthropic says it won't use the data for training
Pricing
This is roughly double the price of Opus 4.8, and the same price as Mythos 5. For developers, that price tag alone may serve as a deterrent for casual use.
Availability
Fable 5 is rolling out in stages:
- Now: Available through the Claude API, consumption-based Enterprise plans, and cloud marketplaces (AWS Bedrock, Google Cloud, Microsoft Foundry)
- Through June 22: Included at no extra cost in Pro, Max, Team, and seat-based Enterprise plans
- June 23 onward: Pulled from standard subscription plans; requires usage credits. Anthropic plans to restore it as a standard subscription feature "as soon as possible"
- Model name on API:
claude-fable-5
What Customers Are Saying
"Fable 5 is a real step forward for the developers GitHub serves. It took on complex, long-horizon coding tasks with a level of autonomy and reliability that exceeded previous benchmarks." — GitHub
"Fable 5's reasoning is a clear step beyond Opus 4.8. It works at senior research scientist grade — picking directions, allocating resources, killing its incorrect beliefs, and producing novel first-principles outputs." — IMC
"Fable 5 is the strongest finance-first model we've tested, both on general finance and reasoning. It's a notable step up." — Jane Street
"What excites us most is the direction it points: a future where developers can hand increasingly ambitious work to agents and trust the results across the software lifecycle." — GitHub
The Bigger Picture: Mythos 5 and Project Glasswing
Alongside Fable 5, Anthropic launched Claude Mythos 5 — the same model with safety restrictions lifted in certain areas. Mythos 5 is deployed through Project Glasswing, a collaboration with the US government that includes partners like Amazon, Apple, Google, Microsoft, NVIDIA, and CrowdStrike.
Mythos 5 has demonstrated:
- 10x acceleration in drug design processes, with 9 of 14 protein targets yielding strong drug design candidates
- Novel scientific hypotheses — preferred by Anthropic scientists over Opus-class models ~80% of the time in blind comparisons
- Autonomous genomics research over a week of work, assembling single-cell data for millions of cells across 138 animal species and training a custom ML model that outperformed a Science-published model despite being 100x smaller
The Controversy: Defanged or Responsible?
Critics have called Fable 5 a "nerfed" or "defanged" version of Mythos — a model that had its claws pulled to make it safe for mass consumption. The mandatory 30-day data retention policy has raised eyebrows, with some viewing it as a precedent for surveillance-like terms attached to powerful AI access.
Anthropic's position is clear: the capabilities are real, the risks are real, and the guardrails are a necessary trade-off to get this technology into the hands of builders and enterprises without waiting for perfect safety. "To release the model both safely and quickly, we've tuned these safeguards conservatively," the company stated.
Should You Use It?
If you're a developer working on complex, long-horizon coding tasks — or an enterprise doing knowledge work that demands sustained reasoning across days — Fable 5 represents a genuine leap forward. The $10/$50 per million token pricing is steep, but for problems that previous models couldn't touch, the value proposition is clear.
For casual use or straightforward tasks, Opus 4.8 remains a more economical choice.
Sources: Anthropic official announcement, CNBC, TechCrunch, ZDNet, CNET, NBC News, Business Insider, Tom's Hardware, Microsoft Azure Blog, Amazon News, and customer testimonials cited by Anthropic.