Agents & InferenceSimon Willison

Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war

Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.

Match the models (Optional)

Which model wrote which summary? Select a matchup mapping below before voting.

Summary A

GPT-6 Luna just dropped to $0.10/$0.50 per million tokens—half its predecessor and roughly one-tenth of Claude Haiku 4.5—making it one of the cheapest capable models OpenAI has ever shipped, while Opus 5.5 cut prices 20% ($4/$20) and slashed cache reads 60%. If you're running high-volume app backends or long agentic loops, re-benchmark now: the cheap tier just got dramatically cheaper, cached-context agents get materially cheaper on Anthropic, and models like GPT-5.6 Terra no longer have any pricing rationale.