Claude Fable 5:
The Model That Changes Everything
A plain-English breakdown of what it is, what it can do, and why it matters — for builders, creators, and thinkers.
The Vault Just Opened
Picture a vault. Inside it is the most capable AI system ever built by Anthropic. They built it, tested it, and then locked it away — not because it was broken, but because it was too powerful to release without the right precautions.
That vault opened on June 9, 2026.
Claude Fable 5 is a Mythos-class model made safe for general use. Its capabilities exceed those of any model Anthropic has ever made publicly available — state-of-the-art on nearly all tested benchmarks, with exceptional performance in software engineering, knowledge work, vision, and scientific research. The longer and more complex the task, the larger its lead over every other model.
Where Did It Come From? The Mythos Origin
To understand Fable 5, you need to understand Mythos — because Fable 5 is Mythos, adapted for the public.
Mythos is the internal name for Anthropic's most powerful AI tier — a step above the entire Opus line. Earlier in 2026, Anthropic gave a small, vetted group of partners access to Mythos Preview under a program called Project Glasswing. The reason? The model was so skilled at finding and exploiting software vulnerabilities that a wide release seemed genuinely dangerous. Governments were consulted. Cybersecurity agencies were involved.
So how did it become public? Anthropic engineered a smart guardrail layer — a set of classifiers that catch dangerous requests and redirect them. The result: you get all the intelligence, none of the risk.
Five Areas Where Fable 5 Rewrites the Rules
This is where we stop talking abstractions and show real examples. Here are the five domains where Fable 5 is redefining what's possible.
Software Engineering
Stripe reported Fable 5 compressed months of engineering into days. In a 50-million-line Ruby codebase, it performed a complete migration in a single day — a job that would have taken a full team over two months by hand.
SWE-Bench Pro: 80.3%Vision
Fable 5 can rebuild a web app's source code from screenshots alone. It completed Pokémon FireRed using only raw game screenshots — no maps, no aids, no helper tools. Previous Claude models couldn't do this even with assistance.
New state-of-the-artMemory & Long Tasks
Fable 5 stays focused across millions of tokens and improves its own outputs using persistent notes. In testing, memory-enhanced Fable 5 outperformed Opus 4.8 by three times on the same task. The better the model, the more memory amplifies it.
3× memory advantageKnowledge Work
On Hebbia's Finance Benchmark for senior-level reasoning, Fable 5 scored highest of any model, with gains in document reasoning, chart interpretation, and complex problem solving. Trading firm IMC said it aced their evaluations nearly across the board.
#1 Finance BenchmarkScience — Generating New Hypotheses
This is where things get extraordinary. Mythos 5 (Fable's restricted sibling) is the first Claude model to consistently produce novel scientific hypotheses. In blinded comparisons, scientists preferred its molecular biology hypotheses roughly 80% of the time. One hypothesis — a novel mechanism for an E. coli protein — was independently corroborated by a separate research lab that was working on the same problem. An AI generated a hypothesis that turned out to be correct. That is genuinely new territory.
~80% scientist preference · Independent corroboration confirmedSWE-Bench Pro · Agentic Coding Benchmark
The Guardrails: Why They Make This Possible
Fable 5's safety layer isn't a limitation — it's the very reason this model exists publicly at all. Without it, Mythos stays locked.
How the Guardrails Work
If you ask Fable 5 about cybersecurity exploits, offensive biology, chemistry synthesis, or AI model distillation, the system automatically redirects your query to Claude Opus 4.8 and notifies you. The filters trigger in less than 5% of sessions on average — so for content creation, research, coding, ministry work, or business documents, you will almost never notice them.
How Do You Start Using Fable 5?
Right now, Fable 5 is available through several channels depending on how you want to use it.
Claude.ai
The simplest way. Visit claude.ai, sign up or log in, and select Fable 5 from the model picker. A paid plan (Pro, Max, Team, or Enterprise) is required to access it.
Claude API
For developers building products. Use the model ID claude-fable-5 via Anthropic's API. Pricing is $10 per million input tokens and $50 per million output tokens.
Cloud Platforms
Also available on Amazon Bedrock and Google Cloud Vertex AI for teams already building within those ecosystems — no separate Anthropic account needed.
Already a Paid Subscriber?
If you're on a Pro, Max, Team, or Enterprise plan, Fable 5 is included at no extra cost for the duration of your current subscription period. No upgrade needed — just log in, select Fable 5 from the model picker, and start using it.
The One-Paragraph Summary
Anthropic spent years building an AI so powerful they were afraid to release it. They finally solved the safety problem well enough to let the public in — with one rule: ask it to help break things, and it politely steps aside. What you get is a model that codes like a senior engineer, sees like a scientist, remembers like an archivist, and reasons like a partner who never gets tired. You have until June 22 to use it free. After that, it costs more — but for hard, complex, long-horizon work, nothing publicly available comes close.