Release Information
Anthropic announced the launch of Claude Fable 5 and Claude Mythos 5 on June 9, 2026, US time. Among them, Claude Fable 5 is positioned as Anthropic's "most capable widely available model" to date, open to general users and developers. Claude Mythos 5 shares the underlying model capabilities with Fable 5 but has fewer safety limitations on certain topics, and is currently only available to approved cybersecurity defenders and infrastructure partners in Project Glasswing.
Claude Fable 5 is positioned by Anthropic as a Mythos-class model targeting the highest capability demands, with a capability tier higher than the existing Opus 4.8. It is primarily oriented towards challenging scenarios such as complex reasoning, long-horizon agentic tasks, software engineering, knowledge work, visual understanding, and scientific research. The company states that the longer and more complex the task, the more pronounced Fable 5's advantages are compared to previous Claude models.
Model Benchmark Scores
Anthropic has published the benchmark scores for Claude Fable 5 and Claude Mythos 5, demonstrating all-around advantages:
Model Information
For developer capabilities, the Claude API documentation shows that the API model ID for Fable 5 is claude-fable-5, supporting a 1M token context window, up to 128K token output, and enabling adaptive thinking by default. The model has been officially available since June 9, 2026, across the Claude API, Claude Platform on AWS, Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.
Pricing comparison can be found in the table below. The standard API prices for Claude Fable 5 and Claude Mythos 5 are approximately double those of Claude Opus 4.8; Fable 5 pricing is halved for batch processing scenarios.
| Model | Standard Input Price | Standard Output Price | Batch Input Price | Batch Output Price |
|---|---|---|---|---|
| Claude Fable 5 | $10 | $50 | $5 | $25 |
| Claude Mythos 5 | $10 | $50 | Not specified | Not specified |
| Claude Opus 4.8 | $5 | $25 | — | — |
Table prices are based on rates per 1 million tokens, reference: Model Pricing - Claude API Docs
Limitations
Built-in Request Moderation
The release of Claude Fable 5 comes with more stringent safety mechanisms. Anthropic stated that if the model detects requests involving high-risk topics such as cybersecurity, biology, chemistry, or model distillation, Claude Fable 5 may not answer directly but instead fall back to Claude Opus 4.8 or return a refusal. The company notes that the safety classifier was intentionally tuned conservatively during the early rollout phase, which might trigger false positives on some harmless requests. However, Anthropic says that over 95% of Fable sessions will not trigger a fallback.
Usage Restrictions
Anthropic states that Claude Fable 5 will be included in Pro, Max, Team, and seat-based Enterprise plans between June 9 and June 22. Starting June 23, unless capacity allows for an extension, continued use of Claude Fable 5 will require usage credits or API calls, meaning additional payments will be required to use it.
Data Retention & Privacy Restrictions
Because Anthropic introduced stronger security monitoring mechanisms in the new models, there are also limitations on how long usage data is retained. The Claude API documentation shows that Claude Fable 5 and Claude Mythos 5 are listed as Covered Models, requiring usage data to be retained for 30 days, and do not support Zero Data Retention (ZDR) arrangements. Requests from organizations that do not comply with this requirement will return a 400 invalid_request_error.
How to Use
The Claude Fable 5 model is immediately available on Anthropic's web interface, API, and Claude Code. Claude Code needs to be updated to version v2.1.170 or above. If Claude Code does not update automatically when opened, you can run the following command to update it manually:
1 | claude update |
Then run /model to switch. The info page already displays that the estimated usage consumption is 2 times that of the Opus model.
Release Perks
To celebrate the release of Claude Fable 5, Anthropic has temporarily reset users' Claude usage limits, including the 5-hour usage limit and the weekly usage limit, allowing users to get more hands-on experience during the new model's initial rollout.
Summary
Overall, the release of Claude Fable 5 marks the beginning of Anthropic pushing the Mythos-class capabilities, previously validated only in controlled projects, to the mass market. Its main selling point is not speed or low cost, but rather a higher capability ceiling in long contexts, complex code, knowledge work, and visual tasks. However, at the same time, the higher API pricing, safety fallback mechanisms, temporary availability on subscription tiers, and data retention schedules associated with safety monitoring mean that enterprises and power developers need to re-evaluate costs, compliance, and product experience before adopting it.
Comments