# Anthropic's Fable 5.2 and Mythos 5.2 arrive in llm-anthropic 0.31

published: 2026-11-10T10:05:27.856Z
topic: infrastructure
source: https://simonwillison.net/2026/Nov/9/llm-anthropic/
url: https://aiagentnews.io/stories/anthropic-s-fable-5-2-and-mythos-5-2-arrive-in-llm-anthropic-0-31

The llm-anthropic 0.31 plugin for the LLM command-line tool adds support for Anthropic's newly released Fable 5.2 and Mythos 5.2 models. Fable 5.2 is priced at $8 per million input tokens and $40 per million output tokens.

Anthropic released Fable 5.2 and Mythos 5.2 on the same day, and Simon Willison has added both to llm-anthropic, the plugin that lets the LLM tool call Claude models. After installing the plugin and storing an Anthropic key, developers can run prompts with the claude-fable-5.2 model identifier. Fable 5.2 is currently ranked first on Artificial Analysis.

Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, a 20 percent discount on Fable 5.1, which cost $10 and $50. Willison tested the model at five reasoning levels, from low to max. At max effort, the model used 128,000 thinking tokens and ran out before returning an image.

Willison says the same max-effort failure appeared earlier with Opus 5.5, but it is new to the Fable series, since Fable 5 and 5.1 returned max-effort output without problems. He also reports using Fable 5.2 to help identify the source of a beep.

## Why it matters if you build agents

Agents that call Fable 5.2 at max reasoning effort can exhaust their token budget without getting any output, so developers should test lower effort levels or handle empty results.

The lower per-token price than Fable 5.1 reduces the cost of running the same workloads on the new model.