Anthropic releases Fable 5.2 and Mythos 5.2, and in its own FrontierCode 2.0 run Fable 5.2 matches Opus 5.5's best score at about half the cost per task

Concept illustration of a building with a wide shop window under a black sign lettered ANTHROPIC, with an orange angular shape to the left of the word; there is a row of windows on the floor above and a row of windows and a door on the floor below. Inside the window, on one shelf from left to right, are a tall stack of cream coins and two laptops labelled OPUS 5.5 and FABLE 5.2, each sitting on a stack of papers with grey text lines and a speech bubble on its screen, and on the far right a shorter stack of red coins in front of a pale pink strip. Printed on the image: About 1/2 Fable 5.2 cost/task vs Opus 5.5, Anthropic test.

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9 (Anthropic, 2026-11-09). In Anthropic's tests, Fable 5.2 at its High reasoning-effort setting matches Opus 5.5's best score at about half the cost per task, a gap list prices alone don't show.

Fable 5.2 and Mythos 5.2 share one underlying model, Anthropic states. Fable 5.2 is open to all users and carries added cybersecurity and biology safeguards that block most security tasks. Mythos 5.2 drops some of those safeguards and goes only to vetted organizations. Both have the same list price, US$8 per million input tokens and US$40 per million output tokens. Anthropic says that is lower than the previous generation. The release page doesn't give the previous generation's price.

The per-token price doesn't show the cost gap. It shows up when scores are compared at each reasoning-effort setting. FrontierCode 2.0 measures whether open-source maintainers would merge the code changes a model writes. Fable 5.2 at High effort scored 45.3%. Opus 5.5's best score across its effort settings was 45.5%. Fable 5.2's cost per task was about half of what Opus 5.5 spent on that 45.5% run. Fable 5.2's 45.3% at High also tops the best score of OpenAI's GPT-6 Astra, 42.8%, at about the same cost per task. At Xhigh effort Fable 5.2 scored 49.1%, against 32.4% for the previous Fable 5.1 at the same setting. Anthropic still recommends Opus 5.5 first for most coding and knowledge work, and keeps Fable for the most ambitious, open-ended technical work.

Anthropic held back, on purpose, the capabilities that most directly speed up AI development. The release page says that in related areas, including accelerator kernel optimization and improving model architectures, Fable 5.2 sits at about Opus 5.5's level.

Not proven yet: the FrontierCode scores and cost ratio are Anthropic's own; no outside rerun is attached How we checked →

Next to watch: whether outside evaluators or public leaderboards that rerun Fable 5.2 also find a cost gap of about half at equal scores.

Source and checks

This item ran in the brief of Nov 10, 2026, alongside 12 other items. How we checked it, the quote checks and what was still missing that day are on that issue's How we checked page.

Cite this item

SecondSource (2026). "Anthropic releases Fable 5.2 and Mythos 5.2, and in its own FrontierCode 2.0 run Fable 5.2 matches Opus 5.5's best score at about half the cost per task." https://secondsource.io/en/news/2026-11-10-1/. Accessed access date.