Anthropic has launched Claude Fable 5.2 and Claude Mythos 5.2, the latest versions of its advanced AI models. The new Claude models bring improvements in coding, reasoning, scientific research, and visual understanding. They also generate responses up to 30% faster than their previous versions.
Anthropic has also cut API prices by 20%, making its latest AI models cheaper for developers and businesses. However, access to Claude Mythos 5.2 remains restricted to verified organizations working in cybersecurity and life sciences.
The company announced both models on November 9, 2026. The launch comes as competition in the AI industry grows, with major technology companies investing heavily in advanced AI models, coding tools, and AI agents.
Claude Fable 5.2 Gets 30% Faster Performance and 20% Lower API Prices
Claude Fable 5.2 and Claude Mythos 5.2 use the same underlying AI model. However, Anthropic has introduced different safety restrictions and access limits for each version.
According to Anthropic, both models deliver responses up to 30% faster than their 5.1 versions. They also provide clearer responses and use fewer tokens for some complex tasks. Anthropic has also reduced the Claude Fable 5.2 API pricing by 20%, making the models less expensive for developers and businesses.
The updated API pricing is:
| Input Tokens | $8 per million tokens, down from $10 |
| Output Tokens | $40 per million tokens, down from $50 |
Anthropic has also added an optional Fast Mode for developers who need faster responses. This feature increases token generation speeds by up to 2.5 times. However, it comes at twice the standard API price.
The Fast Mode pricing is:
| Input Tokens | $16 per million tokens |
| Output Tokens | $80 per million tokens |
The lower API prices could make Claude 5.2 models more affordable for businesses and developers. Faster response times could also help companies handle complex coding, research, and other demanding AI tasks more efficiently.
ALSO READ: Anthropic Launches Claude Opus 5.5 With Fable-Level Performance at 60% Lower Cost
Claude Mythos 5.2 Access Limited to Verified Organizations
While Claude Fable 5.2 is available to a wider range of users, Anthropic has restricted access to Claude Mythos 5.2. The company is making the more powerful version available only to verified organizations.
Both models share the same underlying AI capabilities. However, Claude Mythos 5.2 has fewer restrictions on cybersecurity and biological research tasks. Organizations must pass Anthropic’s verification process to access the model.
Cybersecurity organizations can apply through the Cyber Verification Program. Life sciences researchers can request access through the Life Sciences Verification Program. Anthropic has introduced these restrictions to reduce the risk of AI misuse.
Advanced AI models could help users identify software vulnerabilities or conduct complex biological research. These capabilities could also create security risks if used for harmful purposes.
The company has also improved the safety filters in Claude Fable 5.2. According to Anthropic, its safety classifiers produce around 55% fewer unnecessary blocks than those in Fable 5.1. The improvements include:
- Cybersecurity: Around 60% fewer unnecessary blocks.
- Biology: Around 40% fewer unnecessary blocks.
These changes allow developers to complete legitimate technical and research tasks with fewer interruptions while keeping safeguards in place.
Claude Fable 5.2 Shows Major Improvements in Coding and Advanced Reasoning
Anthropic says Claude Fable 5.2 and Claude Mythos 5.2 deliver major improvements in software development, scientific research, and complex problem-solving. During early testing, both AI models helped engineers complete complex software migrations.
They also designed three working electronic devices from scratch and supported research into robotics. Anthropic also reported that the models solved two previously unsolved problems in black hole physics.
In another test, Claude Fable 5.2 helped researchers formalize a proof of Fermat’s Last Theorem in just 30 hours. The task involved using Lean, a mathematical programming language, to convert the proof into a format that computers could verify.
The new Claude models also showed improvements in agentic coding and advanced reasoning. Anthropic reported the following benchmark results:
- FrontierCode 2.0: Claude Fable 5.2 scored 49.1% at its highest reasoning level, compared with 45.5% for Claude Opus 5.5.
- FrontierSWE v2: Fable 5.2 achieved 69.9%, while Opus 5.5 scored 62.3%.
These benchmark results show Claude Fable 5.2’s improved ability to handle complex coding and software engineering tasks. Its progress in mathematical reasoning and scientific research also points to broader improvements in advanced AI capabilities.
ALSO READ: Anthropic Unveils Claude Haiku 5.5 for High-Volume AI Tasks
Anthropic Expands Claude Fable 5.2 Access Across Major Cloud Platforms
Anthropic has made Claude Fable 5.2 available across its platforms, including the Claude API and major cloud providers such as Amazon Web Services (AWS), Google Cloud, and Microsoft Azure.
Developers can access the new AI model through the Claude Platform by using the model name claude-fable-5-2. Anthropic is also offering eligible paid subscribers additional Claude Fable 5.2 usage during the launch week. The extra usage will depend on their subscription plans.
However, Claude Mythos 5.2 remains available only to verified organizations. Anthropic has also introduced access through Claude Security for eligible Enterprise customers. The launch reflects Anthropic’s efforts to make its latest AI models faster and more affordable.
The company is keeping stricter access controls on models with advanced cybersecurity and biological research capabilities. As competition in the AI market increases, Anthropic is taking a different approach to model availability. It is offering wider access to Claude Fable 5.2 while limiting access to Mythos 5.2 because of its more sensitive capabilities.
