On September 22, OpenAI introduced GPT-6 Sol and GPT-6 Luna, two new models that extend the GPT-6 family beyond GPT-6 Astra. They inherit Astra’s strengths in professional work, factuality, coding, computer use and alignment, but are optimized to be faster and more affordable for large-scale use.
Capabilities & pricing
Both models are reasoning models that accept text and image inputs and produce text outputs via the Responses and Chat Completions APIs.
Key specs for GPT-6 Sol:
- Targeted at complex coding and agentic workflows
- 1,050,000-token context window
- Up to 128,000 output tokens
- Knowledge cutoff: April 20, 2026
- Reasoning effort levels:
none,low,medium(default),high,xhigh,max
Key specs for GPT-6 Luna:
- Optimized as OpenAI’s most efficient model for focused, high-volume tasks
- 1,050,000-token context window
- Up to 128,000 output tokens
- Knowledge cutoff: May 18, 2026
Standard API pricing per 1M tokens (for prompts up to 272K input tokens):
- GPT-6 Sol: $2 input, $0.20 cached input, $10 output
- GPT-6 Luna: $0.10 input, $0.01 cached input, $0.50 output
For comparison, previous GPT-5.6 promotional pricing was: Sol $4 input / $20 output; Luna $0.20 input / $1.20 output. OpenAI says caching and inference are now more efficient, and the savings are being passed on via roughly 50% lower prices.
For both models, Batch and Flex processing are priced at 50% of Standard. Prompts with more than 272K input tokens are billed at 2× input and 1.5× output.
Quality and availability
On OpenAI’s internal factuality evaluations, GPT-6 Sol makes about half as many mistakes as GPT-5.6 Sol. Both Sol and Luna also improve alignment compared to their GPT-5.6 counterparts.
GPT-6 Sol and Luna are rolling out in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users, and are also available via the API. Free and Go users can try GPT-6 Luna in the desktop app.
On September 29, 2026, OpenAI additionally released GPT-6.1 Sol, a newer Sol model for complex coding and professional work at the same $2 input / $10 output pricing; the GPT-6 Sol model page now points to it.
Source: OpenAI
