Anthropic Launches Claude Fable 5, a Safer Mythos-Class AI

Anthropic has officially launched Claude Fable 5, its first publicly available model within the Mythos-class tier of artificial intelligence, as announced on June 9, 2026. Claude Fable 5 is being positioned as Anthropic’s most advanced AI offering for general use, delivering state-of-the-art performance across complex technical, scientific, and analytical tasks while incorporating stringent safeguards to mitigate risks of misuse.

Mythos-class models, until now, have been restricted to trusted partners under Anthropic’s Project Glasswing, which collaborates with the U.S. government and critical infrastructure providers. With this launch, Anthropic has expanded access to the general public for the first time, albeit with safety measures in place that redirect certain high-risk queries—such as those involving cybersecurity or advanced biology—to its less-capable Claude Opus 4.8 model. According to Anthropic, these safeguards are designed to balance operational readiness with responsible deployment.

Breaking Down Claude Fable 5’s Capabilities

Anthropic claims that Claude Fable 5 outperforms all previous Claude models in nearly every benchmark. Its ability to handle long and complex tasks, such as multi-step reasoning and persistent software projects, sets it apart. For instance, during testing, Fable 5 completed a 50-million-line Ruby codebase migration in a single day—a task that would typically take a software team over two months.

Beyond software, its performance in knowledge work is also notable. On Hebbia’s Finance Benchmark, which evaluates senior-level reasoning, Fable 5 achieved the highest score of any AI model yet, excelling in areas like financial analysis, root-cause problem solving, and conceptual reasoning. Furthermore, its multimodal capabilities enable advanced tasks, such as rebuilding web apps from screenshots and analyzing complex scientific visual data.

In life sciences, the restricted Claude Mythos 5 model—essentially an unrestricted version of Fable 5—has reportedly accelerated drug discovery processes by up to tenfold during Anthropic’s internal tests. The model has also produced novel and experimentally validated scientific hypotheses in molecular biology, highlighting its potential to advance research.

Pricing and Access

Claude Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, making it Anthropic’s most expensive general-access model to date. This pricing reflects its increased capabilities and operational scale, including support for a 1 million token context window and up to 128,000 output tokens per request.

For now, public access to Fable 5 is fully available through Anthropic’s API and enterprise plans. However, Anthropic is taking a staged approach to broader rollout, with the model offered at no additional cost to Pro, Max, and enterprise subscription plans until June 22. Starting June 23, usage credits will be required for continued access, reflecting anticipated high demand.

Safeguards and Implications

While Fable 5’s capabilities are groundbreaking, Anthropic has emphasized the importance of its safeguards. Safety classifiers prevent the model from engaging in certain risky queries, such as those related to hacking or dual-use biology research, redirecting such requests to Opus 4.8. According to the company, this safety protocol triggers in less than 5% of sessions, striking a balance between accessibility and ethical considerations.

However, the restricted Claude Mythos 5 model, which has lifted safeguards in certain areas, remains available only to Project Glasswing participants and a small group of life sciences researchers under a trusted-access program. This limited distribution reflects Anthropic’s cautious stance on the deployment of frontier AI capabilities that could be misused.

Market Context and Future Outlook

The release of Claude Fable 5 marks a significant step in the AI industry, where Anthropic is vying with competitors like OpenAI and Google for leadership in next-generation models. By combining cutting-edge performance with a robust safety framework, Anthropic is positioning itself as a responsible innovator in the rapidly advancing field of AI.

As more capable models emerge, Anthropic’s ability to enforce safeguards and ensure ethical use will likely influence adoption rates among enterprise and public users. For now, the launch of Claude Fable 5 serves as both a technological milestone and a case study in navigating the risks of democratizing advanced AI.

Image source: Shutterstock