TL;DR
Anthropic released Claude Fable 5 on June 9, 2026, calling it its most capable model made generally available. The model uses safety classifiers that route flagged cybersecurity, biology and chemistry prompts to Claude Opus 4.8 instead of refusing them, while the less restricted Mythos 5 remains limited to trusted partners.
Anthropic released Claude Fable 5 on June 9, 2026, making what it calls its most capable public model available while reserving a less restricted version, Claude Mythos 5, for trusted partners. The launch matters because Anthropic is pairing higher model capability with a new release design: flagged risky queries are routed to Claude Opus 4.8 rather than answered by the stronger model or simply refused.
According to the source material, Fable 5 and Mythos 5 are the same underlying model. Fable 5 is the public, safeguarded version available through the API as claude-fable-5. Mythos 5 is described as the same model with some safeguards lifted, initially available to Project Glasswing cyber-defenders and expected later for select biology researchers.
The safety system is central to the release. Anthropic’s classifiers monitor areas including cybersecurity, biology, chemistry and distillation. The source material says more than 95% of queries are handled directly by Fable 5, while fewer than 5% are routed to Opus 4.8. Users are told when a fallback occurs. Anthropic’s reported approach means some legitimate security or life-science work may be handled by the weaker model if the classifier is conservative.
The company and outside reviewers reported large gains in coding and long-horizon tasks. Every, which had a week of early access, rated Fable 5 at 91 out of 100 on its Senior Engineer benchmark, compared with 63 for Opus 4.8 and 62 for GPT-5.5. The source material also cites Anthropic-reported examples including a Stripe codebase migration across a 50 million-line Ruby codebase completed in one day rather than two months by a team, drug-design acceleration of about 10 times with Mythos 5, and a genomics model trained by Mythos 5 that beat a recent Science result at one-hundredth the size. Those figures are reported by the named sources and have not been independently verified here.
Builders Face A New Model Split
The release gives developers access to a model Anthropic places above its Opus line, but not with the same permissions for every task. For businesses using frontier models, that changes the buying decision from raw performance alone to a mix of speed, cost, safety routing and operational fit.
The reported pricing is $10 per million input tokens and $50 per million output tokens, less than half the listed price of Mythos Preview but about twice Opus 4.8. The source material says Fable 5 will be free on Pro, Max, Team and Enterprise plans through June 22, then move to usage credits before standard billing later. That staged rollout suggests Anthropic expects demand and supply constraints to shape access.
The model also appears aimed at a narrower type of work than everyday chat. Every described it as strongest on hard, long-running engineering tasks, while warning that lower-adoption users struggled to find a use and that it can be slow and token-heavy. For teams, the practical shift is toward assigning larger jobs to agents, reviewing outputs carefully and using cheaper models for routine edits.
AI VoiceWriter – Smart Dictation & AI Writing Assistant for Windows & Mac | USB Dongle & Mobile App for Voice Input, Proofreading, Rewriting & Multilingual Support
🎙️ Hands-Free Voice Typing for Windows & Mac – Powered by iOS & Android dictation technology, AI VoiceWriter…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Opus To Mythos Class
Anthropic has positioned Fable 5 as the first public model in its Mythos class, a tier the source material says sits above Opus. The company had previously judged this class too risky for broad release, according to the supplied material, which makes the safety routing mechanism part of the product story rather than an add-on.
The naming reflects the split. Anthropic is described as linking “Fable” to the Latin fabula, meaning “that which is told,” and to the Greek mythos. In product terms, the two names mark different access rules for the same underlying model: Fable is public and safeguarded, while Mythos is more open but restricted.
Project Glasswing, Anthropic’s cyber-defense program run with the U.S. government, is listed as the first deployment channel for Mythos 5. The source material says select biology researchers are expected to get access later, but does not provide dates or eligibility details.
“the most capable model it has ever made generally available”
— Anthropic, according to the supplied launch material
AI cybersecurity prompt routing tool
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Safety Routing Has Open Questions
It is not yet clear how often legitimate cybersecurity, biology or chemistry work will be routed away from Fable 5 in real use. The supplied material says flagged prompts make up less than 5% of queries, but also says the classifiers are tuned conservatively and may catch benign requests.
Several reported performance claims remain source-attributed rather than independently confirmed here, including the Stripe migration example, the drug-design acceleration figure and the genomics result. Access terms for Mythos 5 beyond Project Glasswing and future select biology researchers are also still limited in the material provided.
Compliance details may matter for enterprise buyers. The source material flags 30-day retention as a question, but does not provide enough detail to determine how it applies across plans, API usage or restricted partner access.
AI model safety classifier software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Rollout Moves Past Free Access
The next milestone is June 22, 2026, when the free access period for Pro, Max, Team and Enterprise users is set to end. After that, Anthropic plans a usage-credit phase before moving Fable 5 to standard pricing.
Developers and companies are likely to test whether Fable 5’s gains on long-horizon tasks justify higher cost and slower runs compared with Opus 4.8 and faster lower-priced models. For restricted work, the key question will be whether Anthropic expands Mythos 5 access beyond Project Glasswing and named research partners.
MLOps for Biomedical Research with Python: Building Scalable, Reproducible, and Production Grade Pipelines for Genomics, Imaging, and Clinical AI Systems
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did Anthropic announce?
Anthropic announced Claude Fable 5, its most capable model made generally available, along with a restricted counterpart called Claude Mythos 5 for trusted partners.
Are Fable 5 and Mythos 5 different models?
The supplied material says they are the same underlying model. The difference is access and safeguards: Fable 5 is public with classifiers active, while Mythos 5 has some safeguards lifted for approved partners.
What happens when Fable 5 sees a risky prompt?
According to the source material, flagged prompts are routed to Claude Opus 4.8 rather than handled by Fable 5. Users are told when this fallback happens.
How strong is Fable 5 at coding?
Every rated it 91 out of 100 on its Senior Engineer benchmark and called it the best coding model it had tested. That is an attributed reviewer assessment, not a universal benchmark result.
What is still unknown?
Open questions include real-world fallback rates for legitimate technical work, broader Mythos 5 access, enterprise retention details and independent verification of several reported performance examples.
Source: Thorsten Meyer AI
