OpenAI has released GPT-6 Sol and GPT-6 Luna—two more affordable models trained using methods similar to those for GPT-6 Astra. Sol is priced at $2 per million input tokens and $10 per million output tokens, while Luna costs — $0,10 and $0,50 respectively.
Both models are now available via API, ChatGPT Work, and Codex. The description also mentions improved request caching for long-running workflows.
For users, this means the introduction of two pricing tiers within a single lineup: Sol is geared toward higher-cost usage scenarios, while Luna focuses on minimizing expenses. Available materials do not allow for an independent comparison of their performance or verification of the claimed benchmarks.
editorial commentary
Why it matters
The likely consequence is an expanded choice for users who value the balance between capabilities and costs. The next observable signal will be independent comparisons of quality, speed, and actual operating costs. Significant uncertainty remains because current information is based on brief metadata rather than full text and verified tests.