Anthropic releases updated Claude Fable and Mythos models
Anthropic announced upgrades to Claude Fable and Claude Mythos on November 9, 2026, describing them as models for demanding technical challenges. CNBC reported this was the company’s first frontier-model launch since CEO Dario Amodei called for slowing AI development in September.

Key takeaways · 4
- 01
Compare the listed standard price of $8 per million input tokens and $40 per million output tokens with the optional fast mode’s $16 and $80 rates.
- 02
Fable 5.2 is available to all users; organizations seeking Mythos must be verified through Anthropic’s programs.
- 03
Anthropic recommends Claude Opus 5.5 as the starting model for most coding and knowledge tasks.
- 04
Treat Anthropic’s FrontierCode results as its own reported testing, not a guarantee for a particular team’s workload.
Capability, speed and pricing
Anthropic says it upgraded Claude Fable and Claude Mythos for demanding technical challenges, with improvements in coding, reasoning and vision.[1] The company says the new models generate outputs up to 30% faster than Fable 5.1 and Mythos 5.1.[1] Anthropic lists prices of $8 per million input tokens and $40 per million output tokens.[1] An optional fast mode, which Anthropic says can speed token generation by up to 2.5 times, costs $16 per million input tokens and $80 per million output tokens.[1] CNBC reports the new models cost 20% less than the 5.1 generation.[2]
Access and safeguards
Anthropic says Fable has cybersecurity and biology safeguards that allow it to be available to all users.[1] The company says Fable 5.2’s safeguards are as robust as Fable 5.1’s while causing fewer unnecessary blocks.[1] Mythos has a narrower access path: Anthropic says organizations pursuing beneficial cybersecurity or biomedical research can apply through its verification programs, and says Mythos 5.2 is available only to organizations verified through those programs.[1][3]
Reported performance and research
Anthropic says the models helped its engineers complete software migrations it considered otherwise infeasible, designed three working electronic devices from scratch, and solved two open problems in black-hole physics.[1] It also says the models formalized a proof of Fermat’s Last Theorem in 30 hours.[1] In Anthropic’s FrontierCode 2.0 testing, Fable 5.2 scored 49.1% at Xhigh effort, the highest score among released models the company tested.[1] In early testing, Anthropic says Fable 5.2 at High effort scored about level with Opus 5.5 at Max effort for roughly half the cost per task.[1]
Launch amid IPO preparations
CNBC reported that Amodei called for slowing the pace of AI development in September, and that this release was Anthropic’s first frontier-model launch since then.[2] Anthropic responded, “We called for pacing, not pausing.”[2] CNBC also reported that Anthropic was preparing for a potentially historic IPO that could take place as soon as the following week, and was aiming for a $1 trillion to $1.5 trillion valuation.[2] Anthropic says it implemented its strongest-ever protections against distillation attacks.[1] The evidence does not establish that an IPO will occur or that the targeted valuation will be achieved.
For teams assessing technical models, the announcement puts price, speed, access limits and reported benchmark results on the table together. Treat the reported results as inputs to a task-specific evaluation, and check access eligibility before planning work around Mythos.
Why it matters
Bring this to your team — one session a day, built for your industry.
kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.
For teamsHow this developed
10 November 2026
Anthropic releases updated Claude Fable and Mythos models