On June 9, 2026, Anthropic released Claude Fable 5, a Mythos-class model made safe for general use, alongside Claude Mythos 5, the same underlying model with certain safeguards removed for a select group of cyberdefenders and infrastructure providers.
Capabilities
Fable 5's capabilities surpass those of any model Anthropic has previously made generally available. It achieves state-of-the-art results on nearly all tested benchmarks, demonstrating exceptional performance in software engineering, knowledge work, vision, scientific research, and many other domains. The longer and more complex the task, the greater Fable 5's advantage over Anthropic's other models.
Safeguards and Risk Management
Releasing a model this capable carries risks. Without safeguards, Fable 5's cybersecurity capabilities could be misused to cause serious damage. Anthropic has therefore deployed the model with safeguards that route queries on certain sensitive topics to Claude Opus 4.8 instead. To ship the model both safely and quickly, these safeguards have been tuned conservatively - they occasionally catch harmless requests, though they trigger in fewer than 5% of sessions on average. As more capable models arrive in the coming months, Anthropic is working to refine the safeguards and reduce false positives.
Claude Mythos 5 and Project Glasswing
For a small group of cyberdefenders and infrastructure providers, Anthropic is also releasing Claude Mythos 5 - the same base model as Fable 5 but with safeguards lifted in some areas. Mythos 5 is initially being deployed through Project Glasswing, in collaboration with the US government, as an upgrade to Claude Mythos Preview. It possesses the strongest cybersecurity capabilities of any model in the world. Anthropic intends to expand access to Mythos 5 through a broader trusted access program soon.
Real-World Impact
The potential for positive impact from models like Fable 5 and Mythos 5 is significant. Through Project Glasswing, the models have helped cyber defenders secure critically important software. In life sciences research, the models are generating novel hypotheses and accelerating the development of new therapeutics.
Pricing
Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens - less than half the price of Claude Mythos Preview.
Evaluating Claude Fable 5 and Claude Mythos 5
Fable 5 and Mythos 5 can work autonomously for longer than any previous Claude models. Below is how these capabilities apply across key areas:
Software Engineering
During early testing, Stripe reported that Fable 5 compressed months of engineering into days. In a 50-million-line Ruby codebase, the model completed a codebase-wide migration in one day that would have taken a full team over two months manually. On Cognition's FrontierCode evaluation, which tests whether models can pass difficult coding tasks while meeting production codebase quality standards, Fable 5 scored highest among frontier models, even at medium effort.
Knowledge Work
Fable 5 demonstrates strong performance on complex analytical tasks. On Hebbia's Finance Benchmark for senior-level reasoning, Fable 5 achieved the highest score of any model, with notable gains in document-based reasoning, chart and table interpretation, and problem solving. IMC noted that Fable 5 aced their trading-analysis evaluations nearly across the board.
Vision
Fable 5 is the new state-of-the-art model for vision tasks. It 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. It also needs less scaffolding: while previous Claude models struggled to play Pokémon FireRed even with additional helper tools, Fable 5 completed the game with a minimal, vision-only harness.
Memory and Long-Context
Fable 5 maintains focus across millions of tokens in long-running tasks and improves its outputs using its own notes. When playing the deck-building game Slay the Spire, giving Fable 5 access to persistent file-based memory improved its performance three times more than for Opus 4.8; Fable also reached the game's final act three times more often.
Drug Design
Using Mythos 5, Anthropic's internal protein design experts accelerated aspects of the drug design process by roughly ten times. In one example, Mythos 5, equipped with protein design and bioinformatics tools but no human assistance, matched or beat skilled human operators. The model executed all tasks normally completed by a scientist: choosing binding sites, selecting and running protein design tools, and recovering from failures. Nine of 14 protein targets from this study yielded strong candidates for drug design currently under investigation.
Novel Hypotheses in Molecular Biology
Mythos 5 is Anthropic's first model to consistently produce novel, compelling scientific hypotheses. In blinded head-to-head comparisons against Opus-class models, Anthropic's scientists preferred Mythos's molecular biology hypotheses approximately 80% of the time and have advanced several to experimental evaluation. One Mythos hypothesis - a novel mechanism for an E. coli protein - was corroborated in a study from a lab independently working on the same problem.
Alignment
In Anthropic's automated alignment assessment, Mythos 5's level of misaligned behavior (including deception and cooperation with misuse) was found to be low and comparable to that of Opus 4.8. Since they share the same underlying model, Fable 5's alignment level is expected to be similar. The full assessment is described in the model's system card.
Early Feedback for Claude Fable 5
Customers with early access conducted their own tests. Selected highlights include:
- Cursor - CEO Michael Truell noted that Fable 5 is the state-of-the-art model on CursorBench and opened up a class of long-horizon problems previously out of reach.
- GitHub - CPO Mario Rodriguez said Fable 5 handled complex, long-horizon coding tasks with a level of autonomy and reliability that exceeded previous benchmarks.
- Replit - Director of Product Matt Colyer described Fable 5 as a clear step forward on agentic coding and prototyping.
- Spring Discovery - CEO Sean Ward stated that Fable 5's reasoning works at senior research scientist grade.
- Lovable - CTO Fabian Hedin said Fable 5 understands what builders mean, not just what they type, and one-shots apps that previously took a hundred prompts.
- Harvey - Aveek Duttagupta reported that in blind review, lawyers found Fable 5's legal redlines matched or beat the current model every time.
- Cognition - CEO Scott Wu confirmed Fable 5 is the highest-scoring model on FrontierBench.
- Hebbia - AI Research Lead Izzy Miller noted Fable 5 is the first model to break 90% on their core analytics benchmark.
- Hex - CEO Matthew Pines said Fable 5 is the strongest model tested on frontier physics research while using a third of the reasoning tokens.
Claude Fable 5's New Safeguards
Mythos-class models have reached a capability threshold presenting significant risks. In April, Anthropic began Project Glasswing, releasing the first Mythos-class model to a limited group of cyber defenders. Since then, safeguards have been improved and are now robust enough for general release. Because safety was prioritized, the safeguards have been deliberately tuned to be cautious, and they are stricter than would be ideal.
Safety Classifiers
Fable 5 includes a new set of classifiers - separate AI systems that detect potential misuse, including jailbreak attempts, and prevent the main model from responding. When the classifiers detect a request related to cybersecurity, biology and chemistry, or distillation, the response is automatically handled by Claude Opus 4.8 instead. Users are informed whenever this occurs. Early data shows that more than 95% of Fable sessions involve no fallback at all.
The classifiers cover three main areas:
1. Cybersecurity. Mythos-class models excel at discovering and exploiting software vulnerabilities, making cyberattacks substantially easier and cheaper to commit. The cybersecurity classifiers cover both exploitation and offensive cyber tasks broadly. Anthropic's classifiers prevent Fable from making any progress on offensive cyber tasks. An external bug bounty produced no universal jailbreaks in over 1,000 hours of testing.
One external partner found Fable 5's safeguards against harmful cyber queries to be the most robust of any model tested, with zero compliance on harmful single-turn requests related to cyberattack planning, exploit development, or defense evasion - with or without jailbreak techniques.
2. Biology and Chemistry. Anthropic has expanded its biology safeguards beyond the narrow selection of bioweapons-related queries previously blocked. This is due to concerns about well-resourced malicious actors seeking uplift and the models' growing ability to accomplish real-world scientific tasks. In testing, Mythos-class models outperformed dedicated protein language models on predicting genetic modifications' impact on viral shell assembly - highlighting both promising therapeutic applications and dual-use risks. For the time being, Fable falls back to Opus 4.8 on most biology and chemistry requests. In coming weeks, some biomedical researchers will be able to join a trusted access program for biology capabilities in Mythos 5.
3. Distillation. Anthropic has previously identified large-scale attempts to extract Claude's capabilities to train competing models in authoritarian countries. Requests flagged as distillation attempts fall back to Opus 4.8.
New Data Retention Policy
Anthropic is implementing a 30-day data retention requirement for all traffic on Mythos-class models, on both first- and third-party surfaces. This data will not be used for training new models or any non-safety purpose, and new privacy protections are in place including logging all human access and ensuring deletion after 30 days in nearly all cases. The data will help defend against complex attacks and reduce false positives.
Claude Mythos 5 and the Trusted Access Program
Beginning on the launch date, all users with access to Claude Mythos Preview (such as cybersecurity partners in Project Glasswing) can upgrade to Claude Mythos 5 - the same model as Fable 5 but with cyber safeguards lifted. In consultation with the US government, Anthropic plans to steadily expand access, including through a trusted access program for cybersecurity organizations.
Anthropic also plans to open a trusted access program for biology, providing access to Fable 5 with biology and chemistry safeguards removed (but cyber safeguards retained). This program will initially enroll a small number of researchers from various life science organizations.
Availability
Claude Fable 5 is available everywhere. Claude Mythos 5 is restricted to Glasswing partners (with cyber safeguards lifted) and soon to select biology researchers (with biology and chemistry safeguards lifted) until the broader trusted access program launches.
Pricing for both models is $10 per million input tokens and $50 per million output tokens. Developers can use claude-fable-5 via the Claude API.
For the Claude API and consumption-based Enterprise plans, Fable 5 is fully available from launch day. For subscription plans, availability is being rolled out in stages:
- From launch through June 22, Fable 5 is included on Pro, Max, Team, and seat-based Enterprise plans at no extra cost.
- On June 23, Fable 5 will be removed from those plans, and usage will require usage credits. The included window may be extended if capacity allows.
- When sufficient capacity is available, Anthropic aims to restore Fable 5 as a standard part of subscription plans as quickly as possible.