Advancing the price-performance frontier with GPT‑5.6
GPT-5.6 Luna’s price cut is large enough that Willison switched his demo off Gemini.
OpenAI cut GPT-5.6 Terra pricing by 20% and Luna by 80%, with the company crediting GPT-5.6 Sol’s inference and kernel optimizations for lower serving costs. Willison highlights Luna’s new $0.20 per million input tokens and $1.20 per million output tokens as cheaper than Gemini 3.1 Flash-Lite on input and far below Claude Haiku 4.5. He says the Luna drop “completely changes the landscape” for lower-priced models. Simon Willison's note
OpenAI cut GPT-5.6 Terra pricing by 20% and Luna by 80%, with the company crediting GPT-5.6 Sol’s inference and kernel optimizations for lower serving costs. Willison highlights Luna’s new $0.20 per million input tokens and $1.20 per million output tokens as cheaper than Gemini 3.1 Flash-Lite on input and far below Claude Haiku 4.5. He says the Luna drop “completely changes the landscape” for lower-priced models. Simon Willison's note
score 7