- OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, 2026, extending GPT-6 Astra’s improvements to faster, cheaper tiers.
- Sol targets complex tasks like coding; Luna is built for high-volume clerical work such as summarizing documents and extracting information.
- OpenAI says GPT-6 Sol makes about half as many mistakes as its GPT-5.6 Sol predecessor, at half the price.
- The models rolled out immediately in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users.
What Happened
OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, 2026, positioning the pair below flagship GPT-6 Astra as faster, lower-cost options, according to TechCrunch’s report from launch day. Both models carry over training methods OpenAI used for Astra, aimed at improving professional work, factuality, coding, and computer use.
Why It Matters
The release continues a pattern for this model line: GPT-5.6 Sol previously autonomously post-trained its own Luna successor, and OpenAI has since kept shipping efficiency gains across the Sol/Luna tier rather than only at the flagship level, as it also did with GPT-6 prompt-caching improvements.
Technical Details
OpenAI positions Sol for complex tasks like coding and Luna for “high-volume tasks with a clear goal,” such as summarizing documents, extracting information, or answering quick questions. OpenAI claims GPT-6 Sol makes about half as many mistakes as GPT-5.6 Sol on its internal benchmark tests, while pricing both new models at half the cost of their 5.6-series predecessors.
Who’s Affected
Developers using Codex and enterprise customers on ChatGPT Work get access first; OpenAI did not include free-tier ChatGPT users in the initial rollout. Cost-sensitive teams that found Astra-tier pricing prohibitive are the stated target for the cheaper Sol/Luna tier.
What’s Next
OpenAI has not said when Sol and Luna will reach free-tier ChatGPT users, following its usual pattern of staging access by subscription tier first.
