GPT-6 Sol and Luna

Two faster, cheaper GPT-6 models trained with the same methods as GPT-6 Astra, at half the API price of the GPT-5.6 versions they replace.

Sol is the mid-tier model for difficult work tasks and Luna is the smallest and cheapest, while GPT-6 Astra stays OpenAI's top model
API prices are $2 input and $10 output per million tokens for Sol (was $4 and $20 for GPT-5.6 Sol) and $0.10 and $0.50 for Luna (was $0.20 and $1.20)
Improved prompt caching gives higher cache hit rates by default, so agents and long conversations that reuse the same context cost less
On OpenAI's AutomationBench test of business workflows across apps, Sol at its highest effort setting beat Claude Opus 5 at max effort for 9% of the cost per task
Rolling out through the day in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, and Free and Go users get Luna in the desktop app
Not yet available in regular ChatGPT chat. In the API they are gpt-6-sol and gpt-6-luna

High-volume everyday work where Astra's full depth is overkill, like business workflows across apps, drafting, and coding at scale.

Cost or usage limits are your constraint. Pick Sol for harder work and Luna for the cheapest, fastest runs.

Paid ChatGPT Work and Codex users, and developers running apps or agents on the OpenAI API.

Join the newsletter to get ahead with AI.

I share strategies, new modules, free guides, and tools to get you better at using AI in your work.