Anthropic's New Claude Model Tops an Independent Ranking, and Costs the Most to Run
Anthropic released two new Claude models on Nov. 9, 2026, and an independent benchmarking firm has put one of them at the top of its table. Artificial Analysis, which measures commercial AI models independently of their makers, scores Claude Fable 5.2 at 65 on its Intelligence Index. That places it first among the 74 comparable models the firm ranks, against a median of 40.
Anthropic says Claude Fable 5.2 and Claude Mythos 5.2 are the same underlying model, built for open-ended technical work rather than everyday use. The company recommends its Opus 5.5 model as the starting point for most coding and knowledge tasks. Fable carries safeguards meant to block cybersecurity and biology misuse and is available to all users. Mythos, which has fewer of them, goes only to organizations accepted into the company's Cyber Verification and Life Sciences Verification programs.
The company prices Fable 5.2 at $8 per million input tokens and $40 per million output tokens, which it says is below the price of Claude Fable 5.1 and Claude Mythos 5.1. It says the new pair also generates output up to 30% faster than those models, and lists Fable 5.2 as available on Amazon Web Services, Google Cloud and Microsoft Azure.
Artificial Analysis also measures what that score costs. On its index, Fable 5.2 is the most expensive of the 73 models it prices, at $26.54 per task, and the second slowest of the 72 it times. All three of its figures describe the model run at its highest effort setting.
On its own benchmarks, Anthropic reports Fable 5.2 scoring 49.1% on FrontierCode 2.0, a test of whether a model's code changes would be accepted into open-source projects, against 45.5% for Opus 5.5. The company also says the model solved two open problems in black-hole physics, for which it has posted two preprints that have not been peer reviewed, and that copies of it working in parallel formalized the proof of Fermat's Last Theorem in 30 hours.
