Claude Fable 5.2 Lands in OpenBot

Fable 5.2 looks most interesting as an agentic-workload upgrade, but Anthropic’s strongest claims remain self-reported until independent evaluations arrive.

• –Adaptive thinking and high-effort modes target ambiguous, multi-step engineering tasks.
• –The 1M-token context window makes large repositories and extended agent sessions more practical.
• –Pricing falls to $8 per million input tokens and $40 per million output tokens, still expensive for heavy autonomous use.
• –OpenBot’s provider switching lets teams test Fable 5.2 against Codex, Gemini, Claude, and local models in the same workspace.
• –OpenBot’s local-first architecture does not eliminate Anthropic’s retention and safety-processing policies for Fable traffic.