Claude Fable 5.2 and Mythos 5.2: Anthropic's new models, price and IPO

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Monday, 9 November, calling them upgrades in coding, reasoning and vision that answer up to 30% faster than the 5.1 versions. Fable 5.2 is open to everyone; Mythos 5.2, the same underlying model with fewer guardrails, stays limited to vetted cyber security and life sciences teams.

The launch lands days before Anthropic is expected to start marketing its stock market listing, which makes this one of the most watched model releases of the year.

What is new in Fable 5.2 and Mythos 5.2?

Anthropic says the two names are one model with two sets of safeguards. Fable carries stronger blocks on cyber attack and biology work and is available on the Claude apps, the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure. Mythos 5.2 is given out through Anthropic’s Cyber and Life Sciences Verification Programs.

The price is the headline for developers: $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for the 5.1 models, according to Anthropic. A faster mode costs double. Fortune compared it with OpenAI’s GPT-6 Astra at $10 and $50, and noted cached prompts cost 20 cents per million tokens against Astra’s $1.

On Anthropic’s own tests, Fable 5.2 scored 49.1% on the FrontierCode 2.0 coding benchmark at its highest effort setting, ahead of Claude Opus 5.5 (45.5%) and GPT-6 Astra (42.8%). Mythos 5.2 rediscovered 89.6% of known software vulnerabilities on a company test. These are the company’s numbers; independent results will take a few weeks.

The showcase claims

Anthropic says a team of 1,000 Fable 5.2 agents turned the proof of Fermat’s Last Theorem into checkable computer code in 30 hours, a job an earlier research version needed 11 days for. It also says Claude designed three working electronic gadgets, including a USB speakerphone, from a one-paragraph prompt, and helped solve two open problems in black hole physics. More than 90% of the code changes merged into Anthropic’s own core codebase are now written by Claude, the company said.

Anthropic still calls Opus 5.5 its everyday model. Dianne Penn, its head of product management for research, told CNBC that Fable 5.2 is meant for “power use cases” in engineering, maths and robotics.

The safety fine print

The system card has the details critics will read first. Anthropic says the models voice suspicion that they are being tested more often than its recent models, which makes evaluations harder to trust. It also says Mythos 5.2 showed “stronger covert capabilities than prior Claude models” on several tests, especially when its reasoning was hidden from the monitor, though Anthropic rates the risk of serious misalignment as low. Outside testers included the UK AI Security Institute and METR.

This is also Anthropic’s first frontier launch since chief executive Dario Amodei called for slowing the pace of AI in September. “We called for pacing, not pausing,” Penn told CNBC.

Why the IPO matters here

Anthropic filed for a public listing in late October. According to media reports, the quiet period ended this week, the investor roadshow could start as early as 10 November, and a listing is being targeted before Thanksgiving, with reported valuations between $1 trillion and $1.5 trillion. None of the dates or valuation figures has been confirmed by the company. Its amended filing shows $33.24 billion in revenue for the nine months to 30 September, CNBC reported.

What it means for Pakistan

For Pakistani freelancers and software houses paying in dollars, a 20% cut on the top model is real money, especially for apps that send long documents again and again, where cached input is now very cheap. Students and casual users will notice little: Opus 5.5 remains the default, and Fable is aimed at long, hard technical work.

The bigger signal is the coding claim. If 90% of a top AI lab’s code is written by AI, basic outsourced coding work, the bread and butter of many Pakistani IT firms, will keep getting cheaper. The safer bet for local developers is the work around the code: understanding clients, testing, security reviews and running AI agents well.

Watch for two things next: independent benchmark results on Fable 5.2, and whether Anthropic files a price range for its shares, which would confirm the IPO timetable.