OpenAI launched GPT-6 Sol and GPT-6 Luna on September 22, expanding the model generation introduced with Astra earlier this month. The release adds two options aimed at different workloads: Sol for complex coding and agentic workflows, and Luna for focused, high-volume tasks.

OpenAI's API documentation lists standard text-token prices of $2 per million input tokens and $10 per million output tokens for Sol. Luna is priced at $0.10 for input and $0.50 for output per million tokens. Cached inputs cost $0.20 and $0.01, respectively; cache writes are billed separately.

Both models accept text and images and produce text. Each lists a 1,050,000-token context window, with a maximum of 922,000 input tokens and 128,000 output tokens. Requests exceeding 272,000 input tokens incur higher rates across the full request, so the headline prices do not apply uniformly to long-context workloads.

For agent developers, both model pages list Responses API support for tools including web search, file search, hosted shell, computer use and MCP. OpenAI directs developers to Responses for built-in tools and function calling; Chat Completions supports function calling only when reasoning effort is set to none.

TechCrunch reports that OpenAI also claims improved factual accuracy over the preceding generation. Those claims come from company evaluations, not independent validation. The concrete change for developers is a broader set of GPT-6 deployment options, with published pricing and tool support that can be tested against their own coding and document-processing workloads.