Anthropic releases Claude Fable 5.2 and Mythos 5.2 with a fast mode

Anthropic released two new models on 9 November. Claude Fable 5.2 is the one most readers will use. Claude Mythos 5.2 is the same model without two safety filters. It goes to cybersecurity and life sciences teams who pass a review. Both are out today on claude.ai, Claude Code and the Claude platform. They are also on Amazon, Google and Microsoft's cloud markets.

What is new

Anthropic says Fable 5.2 generates output up to 30 percent faster than Fable 5.1 at the standard price. The company also opened a new fast mode. Fast mode trades a higher token price for up to 2.5 times the output speed. The standard price is 8 dollars per million input tokens and 40 dollars per million output tokens. Fast mode costs 16 dollars per million input tokens and 80 dollars per million output tokens. Anthropic says fast mode is for agent loops where speed per step changes what the agent can finish.

The scores Anthropic reports

Anthropic published benchmark figures for both models against 5.1 on the same tests. Fable 5.2 reaches 49.1 percent on FrontierCode 2.0 at the highest effort setting. The 5.1 score on the same test was 41.1 percent. On RoboDojo Sim, a robot simulation benchmark, Fable 5.2 scores 64.9 percent against Fable 5.1's 27.6 percent. On BenchCAD 2.1, a computer-aided design test, Fable 5.2 reaches 83.5 against 44.4 for Fable 5.1. On CVE Rediscovery, a security test, Mythos 5.2 reaches 89.6 percent against Mythos 5.1's 78.3 percent. All figures are the vendor's own, measured by Anthropic.

What else the post says

Anthropic says the new models carry what it calls the "strongest-ever protections against distillation attacks". The company's goal is to stop a smaller model being trained from the new one's outputs. A new enterprise safeguard watches for harmful use on accounts that do not retain data. Fast mode also runs on Anthropic's Agent Skills and on Claude Code's built-in subagents.

A team on Fable 5.1 can switch to Fable 5.2 with no code change. The API name is new, but the shape of the request and response is the same. The price per standard token stays the same as 5.1. A team that keeps standard mode gets a higher benchmark score at the same bill. A team on long agent sessions can pay twice the token price for up to 2.5 times the speed. Whether that trade pays off depends on how much of the session time is model time. The rest is tool time, and fast mode will not change it.

Source

• Anthropic, Claude Fable 5.2 and Mythos 5.2
• The Verge, Anthropic launches new Claude Fable and Mythos 5.2 models