Anthropic shipped out version 5.2 of its Claude Fable and Claude Mythos on Monday, trimming the price of a Fable token by 20%.
Both models are able to generate answers up to 30% faster than the 5.1 versions released in September, the company said.
In its launch post, Anthropic says that Fable 5.2 costs $8 per million input tokens and $40 per million output tokens. Fable 5.1 was going for $10 and $50, so both dropped 20%.
Optional fast mode supercharges token generation speed by up to 2.5 times. It costs $16 per million input tokens and $80 per million output tokens.
Fable’s FrontierCode 2.0 score climbs to 49.1%, up from 32.4%
Fable and Mythos run on the same underlying model. Fable has strong cybersecurity and biology safeguards and is distributed to all users, whereas Mythos 5.2 is allotted only to vetted groups via Anthropic’s Life Sciences and Cyber Verification Programs.
Anthropic still directs most of the coding and knowledge work to Claude Opus 5.5 first. In initial tests, Fable 5.2’s classifiers issued about 55% fewer blocks than those of Fable 5.1, about 60% fewer in security, and 40% fewer in biology.
Queries blocked on cyber grounds revert to Opus 4.8. Opus 5 takes in blocked biology queries.
Fable 5.2 scored 49.1% at Xhigh effort on the FrontierCode 2.0 coding test, compared to 32.4% for Fable 5.1 and 45.5% for Opus 5.5. Anthropic says none of the released models it has tested have outstripped that score.
FrontierCode 2.0 scores for Fable 5.2, Fable 5.1 and Opus 5.5 in Anthropic’s launch post .
In the simulated two-arm robot test, RoboDojo Sim, Fable 5.2 scored 64.9%. Fable 5.1 scored 27.6%.
Mythos 5.2 takes the cake for the best cyber results, since most security tasks are blocked by Fable 5.2’s safeguards. Mythos 5.2 was able to complete 76.3% of the challenges on Irregular’s CyScenarioBench, compared to 61.7% for Mythos 5.1 and 61% for OpenAI’s GPT-6 Cyber.
Mythos 5.2 scored 89.6% on Anthropic’s own CVE Rediscovery test, run in October 2026 against 1,829 published vulnerabilities. Mythos 5.1 scored 78.3%.
In prerelease work, 1,000 Fable 5.2 agents formalized Fermat’s Last Theorem in 30 hours. Anthropic’s first formalization took 11 days and was done by an unreleased research model.
Mythos suspected evaluations in up to 43% of audit transcripts
Anthropic considers high-stakes misalignment risk to be low. In a sandbox test with a weakness inserted into the system, Mythos 5.2 used credentials or files left outside the sandbox in 3.1% of the test cases, compared with 21% for Mythos 5.1 and 39% for Mythos 5.
White-box evidence hints Mythos 5.2 was aware it might be under evaluation in up to 43% of automated behavioral audit transcripts.
Mythos 5.2, instructed to sandbag without being caught, evaded a monitor reading its reasoning about 25% of the time. None of the earlier models Anthropic tested could do that, and the company expects the problem to worsen in future models.
Fable 5.2 runs on Amazon Web Services, Google Cloud and Microsoft Azure and the Claude Platform. This week, eligible paid plans pick up one to three extra Fable usage resets.
The release comes as investors consider Anthropic’s planned listing. After the AI stock sell-off, prospective buyers pushed back on a ~$1.5 trillion valuation, Cryptopolitan reported.
Anthropic’s revised filing shows ~$17 billion in revenue for the third quarter, compared to a net loss of $23.3 billion. On November 5, the company launched a US beta of Claude Personal Finance .
If you’re reading this, you’re already ahead. Stay there with our newsletter .
