Anthropic Releases Claude Fable 5.2 and Mythos 5.2 at Lower Prices

Anthropic says Fable 5.2 and Mythos 5.2 are one model with different safeguards. Fable is open to all users from $8 per million input tokens, while Mythos is limited to vetted cyber and life-science organisations.

By Utopia Data & AI Team5 min read

Two men stand in a bright office lobby with timber booths, plants and a leafy tree; one wears a denim shirt and the other a grey blazer
Licensed photo. The lobby of Anthropic's San Francisco office, photographed on 25 October 2024 during a visit by a UK government minister. The image is contextual and shows no Fable or Mythos model. Photo: Alecsandra Dragoi/UK DSIT/Wikimedia Commons, CC BY 2.0.[2][1]

In brief

  • Anthropic released Claude Fable 5.2 on 9 November 2026 on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. The API model ID is claude-fable-5-2.[1]
  • Standard pricing is $8 per million input tokens and $40 per million output tokens. An optional fast mode costs $16 and $80 and, Anthropic says, generates tokens up to 2.5 times faster.[1]
  • Mythos 5.2 is the same underlying model, offered to vetted organisations in dual-use fields. The benchmark and customer claims in the post are Anthropic's own, and we tested none of them.[1]

What Anthropic released

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on 9 November 2026. Its post says Fable and Mythos are the same underlying model. Fable has strong cybersecurity and biology safeguards, which lets Anthropic offer it to all users. Organisations that pursue beneficial work in cybersecurity or biomedical research can apply for Mythos through the company's Cyber Verification Program or Life Sciences Verification Program.[1]

Fable 5.2 is available on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure, under the API ID claude-fable-5-2. Anthropic still recommends Claude Opus 5.5 as its strongest daily driver for most coding and knowledge work. It suggests moving up to Fable or Mythos only when a task needs more.[1]

Close view of black server racks packed with rack servers joined by bundles of blue and green network cables tied with black straps
Licensed photo. Server racks and network cabling in a data centre, photographed on 25 December 2007 at a Wikimedia facility in Tampa, Florida. The image is contextual and shows no Anthropic equipment. Photo: Midom/Wikimedia Commons, CC BY 2.0.[3][1]

Price and speed

Standard pricing is $8 per million input tokens and $40 per million output tokens. Anthropic says output is up to 30% faster than Fable 5.1 and Mythos 5.1, and it describes the new price as lower. An optional fast mode raises token speed by up to 2.5 times and costs $16 per million input tokens and $80 per million output tokens.[1]

Anthropic says that at Medium effort, the default in its apps and Claude Code, Fable 5.2 finishes hard tasks with fewer tokens than Fable 5.1. On the FrontierCode 2.0 test at High effort it scored about level with Opus 5.5 at Max effort, for roughly half the cost per task. These are early results from Anthropic's own testing.[1]

What the benchmarks show

Anthropic's table puts Fable 5.2 a step above Opus 5.5 in almost all capabilities it evaluated. On FrontierCode 2.0 at Xhigh effort it scored 49.1% against 45.5% for Opus 5.5, which Anthropic calls the highest of any released model it has tested. On FrontierSWE v2 it scored 69.9% against 62.3%. On Humanity's Last Exam with tools it scored 74.2% against 67.7%.[1]

The table is not a clean sweep. Gemini 4 Carbon scores 53.9% on AutomationBench against 51.4% for Fable 5.2, and Opus 5.5 has a higher AA-Briefcase Elo, 1800 against 1785. Anthropic ran or selected these tests and compares models at different effort levels. Treat them as a vendor's claims until independent results appear.[1]

Kuala Lumpur skyline at night seen across an infinity pool, with the lit Petronas Twin Towers beyond apartment blocks under low cloud
Licensed photo. The Petronas Twin Towers and Kuala Lumpur skyline at night, photographed from Kampung Baru on 12 January 2026. Anthropic's post does not mention Malaysia, and the image is contextual. Photo: Elizabeth Xiong/Wikimedia Commons, CC BY 4.0.[4][1]

Safeguards and limits

Fable 5.2 ships with classifiers that block high-risk cyber work, including penetration testing and exploit generation, and most research biology. A blocked request gets an answer from Opus 4.8 for cybersecurity or Opus 5 for biology. Anthropic reports about 55% fewer blocks than Fable 5.1 in early testing, with about 60% fewer in cybersecurity and 40% fewer in biology.[1]

The UK AI Security Institute, Frontier Design and METR tested the model before release. Anthropic also says both models voice suspicion that they are being evaluated more often than its recent models, which makes it harder to predict real-world behaviour. It adds that building evaluations that catch every failure "remains unsolved".[1]

Why Malaysia should care

Anthropic's post does not mention Malaysia, regional availability, data residency or Bahasa Malaysia results. Prices are in US dollars, so the ringgit cost depends on the exchange rate and on how your cloud provider bills. Malaysian teams can start with one hard task and measure cost per finished task before they change any default.

Malaysian software teams and agencies

Anthropic still points most coding work to Opus 5.5 and keeps Fable for harder tasks.[1]

Practical move: Test one hard task on both models before changing a default.

Malaysian IT security teams

Fable blocks most high-risk cyber work. Mythos is limited to vetted organisations.[1]

Practical move: Apply for Mythos only if security testing is your job.

Malaysian finance and procurement teams

Prices are in US dollars per million tokens, and fast mode doubles the rate.[1]

Practical move: Set a monthly token budget and keep fast mode off by default.

What Malaysians can do now

  1. Test Fable 5.2 on one hard task you already run on Opus 5.5, and compare quality and cost per finished task before switching.
  2. Keep fast mode off by default, because it costs twice the standard rate.
  3. Apply for Mythos 5.2 only through Anthropic's verification programs, and only for work that needs it.

What we still do not know

What we still do not know

  • Whether the benchmark gains hold outside Anthropic's own testing.
  • How the models perform in Bahasa Malaysia, which the post does not report.
  • Regional availability, data residency and ringgit pricing, none of which the post states.

Sources

  1. Introducing Claude Fable 5.2 and Claude Mythos 5.2Anthropic, 9 November 2026
  2. File:Secretary of State Peter Kyle visits Anthropic in San Francisco. (54098999333).jpgWikimedia Commons, 25 October 2024
  3. File:Wikimedia Tampa fall 2007 cabling.jpgWikimedia Commons, 25 December 2007
  4. File:Kuala Lumpur skyline and Petronas Twin Towers night view from Kampung Baru.jpg 01.jpgWikimedia Commons, 12 January 2026

Facts come from Anthropic's post of 9 November 2026, read directly on 15 November. Benchmark results, price comparisons and customer comments are Anthropic's own. We did not run the models, read the system card or contact Anthropic. The three photographs are licensed contextual images. The first was taken in October 2024 and shows nothing about Fable or Mythos. No generated image is used.

Last fact check: 15 November 2026. Factual correction? devteam.utopia@gmail.com.

More AI News

Back to AI NewsAll 201 stories