Claude Fable 5.2 context window and max output: 1M in, 128K out

Claude Fable 5.2 context window: 1M tokens

Fable 5.2 and Mythos 5.2 take up to 1 million tokens of input, per Anthropic's models overview. Anthropic puts that at roughly 555,000 words or 2.5 million Unicode characters on its current tokenizer, which arrived with Claude Opus 4.7 and produces about 30% more tokens for the same text than older Claude models.

No long-context surcharge

Anthropic bills the full 1M window at the standard rate for Claude 4.6 and later models, Haiku 5.5 excepted, per its pricing page. A 900K-token request costs the same per token as a 9K-token one. That is a real edge over GPT-6 Astra, which charges $20 input and $75 output once a prompt passes 272K tokens.

Example, our math: one 500K-token prompt with a 10K-token answer costs about $4.40 on Fable 5.2 and about $10.75 on GPT-6 Astra at its long-prompt rate.

Max output: 128K tokens

A single synchronous request can return up to 128K output tokens, the same as Fable 5.1 and Opus 5.5. Anthropic's 300K-output beta for the Batch API lists Opus, Sonnet and Haiku models, not Fable 5.2. Adaptive thinking is always on for Fable 5.2, so long reasoning also comes out of that output budget.

On Claude plans

Claude's pricing page lists context "up to 1M, varies by model" on paid plans. Long conversations also burn usage limits faster, because each turn resends the history. See the Claude 5-hour limit guide for how to stretch a session.

Verdict

For very long prompts, Fable 5.2 beats GPT-6 Astra: similar window, no surcharge, and a lower base price. Grok 4.7 is cheaper per token but stops at 500K.

Full launch coverage: Claude Fable 5.2 and Mythos 5.2: prices, limits and what changed.