Announced 9 Nov 2026 · Sources checked
What are the new models and what do they cost?
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Monday, upgrading the models it describes as designed for the most daunting technical challenges.
The company says the new versions show improvements in coding, reasoning, and vision compared to the 5.1 versions released in September.
Claude Fable 5.2 is available to all users on Anthropic platforms, Claude Code, and the API. It costs $8 per million input tokens and $40 per million output tokens, a reduction from $10 and $50 for Fable 5.1.
An optional fast mode at $16 per million input tokens and $80 per million output tokens speeds up token generation by up to 2.5 times.
Claude Mythos 5.2, which shares the same underlying architecture but has weaker safety guardrails for cybersecurity and biological research, is available only to organizations in Anthropic's Life Sciences and Cyber Verification Programs.
The company said Mythos 5.2 is its most capable model for cybersecurity engineering.
How much faster and cheaper is Fable 5.2?
Anthropic says Fable 5.2 and Mythos 5.2 generate outputs up to 30 percent faster than the 5.1 versions across effort levels. Faster inference serving is available for double the price.
Fable 5.2 is now cheaper than its primary competitor, OpenAI GPT-6 Astra. The lower pricing positions Fable 5.2 competitively against Gemini 4 Carbon and Opus 5.5 for high-end tasks while Fable remains Anthropic's most capable model tier.
Fable 5.2 is also compatible with zero data retention policies through the Enterprise Frontier Safeguards system, a feature long requested by enterprise clients according to Anthropic.
What do the benchmarks show?
Anthropic published benchmark results showing Fable 5.2 ahead of GPT-6 Astra and Gemini 4 Carbon on several key evaluations.
On FrontierCode 2.0, a measure of agentic coding, Fable 5.2 at Xhigh effort scored 49.1 percent, compared to 32.4 percent for Fable 5.1 and 45.5 percent for Opus 5.5.
On FrontierSWE v2, Fable 5.2 scored 69.9 percent, ahead of Opus 5.5 at 62.3 percent and GPT-6 Astra at 65.5 percent.
On AutomationBench for business workflows, Fable 5.2 scored 51.4 percent, well above Fable 5.1 at 31.4 percent and Opus 5.5 at 42.5 percent.
For scientific research, Terminal-Bench-Science showed Fable 5.2 at 74.6 percent and Mythos 5.2 at 75.3 percent, compared to Opus 5.5 at 58.7 percent and GPT-6 Astra at 68.1 percent.
On Humanity Last Exam with tools, Fable 5.2 scored 74.2 percent, ahead of Opus 5.5 at 67.7 percent and GPT-6 Astra at 57.2 percent.
What can the models actually do?
According to Anthropic, during prerelease testing, the models helped engineers complete otherwise-infeasible software migrations, designed three working electronic devices from scratch, solved two open problems in the physics of black holes, made substantial progress on robotics evaluations, and formalized the proof of Fermat Last Theorem in 30 hours.
In one experiment, an Anthropic engineer gave Fable 5.2 the task of designing a prototype operating system.
Starting from an empty folder, the model coordinated over 1,000 subagents working around the clock for eight days, producing a working OS that boots on a laptop with drivers, file storage, windows, and built-in Claude functionality.
TechCrunch reported that a startup called Physical Superintelligence announced a new proof of the Aaronson-Ambainis conjecture, a pivotal step in setting theoretical limits for quantum computing, achieved by running a non-public Anthropic model within the startup own system.
Anthropic did not confirm whether Fable 5.2 was the model in question.
How does this relate to RSI concerns?
The release comes a week after a third-party assessment from METR, which suggested Anthropic might be approaching recursive self-improvement thresholds.
found that internal models were speeding up Anthropic AI research progress by roughly 2x, although it noted many uncertainties remain.
Anthropic noted significant ongoing challenges in accurately testing model capabilities.
In the release, the company wrote: Claude Mythos 5.2 and Fable 5.2 each voice suspicion that they are being evaluated more often than any of our recent models, which challenges our ability to assess how they will act in real-world settings.
As these settings expand and capabilities increase with future models, we expect this challenge to grow.
The system card for Mythos 5.2 notes that the model shows stronger covert capabilities than prior Claude models, particularly when its reasoning is hidden from the monitor.
Anthropic maintains that low absolute performance on their covert evaluations supports claims that detection of harmful actions in high-stakes situations remains reliable.
Common questions
How does Fable 5.2 compare to Opus 5.5?
Fable 5.2 is Anthropic most capable model tier, above Opus 5.5. Anthropic recommends Opus 5.5 as a daily driver for most coding and knowledge tasks, switching to Fable or Mythos for the most ambitious technical work.
Fable 5.2 can also serve as an advisor or orchestrator that delegates simpler tasks to smaller models.
Who can access Mythos 5.2?
Mythos 5.2 is only available to verified organizations in Anthropic Life Sciences and Cyber Verification Programs. Fable 5.2 has cybersecurity and biology safeguards that block most misuse-related tasks, which allows Anthropic to make it available to all users.
Sources & further reading
How this story was made
Written by Kristian Kostov with AI assistance and checked against the linked sources. Company performance claims are attributed to the company. Analysis reflects AiLookout’s interpretation; we have not independently tested the products discussed. The cover is an AI-generated editorial illustration, not a photograph or evidence of an actual event.
Our editorial standards
