Launch · AI Assistants
GPT-6 Sol and GPT-6 Luna: what OpenAI launched, what it costs and who gets it
OpenAI's GPT-6 Sol and Luna cost half as much as GPT-5.6 in the API and are live in ChatGPT Work and Codex on Plus and above, but not yet in Chat. See prices and limits.
The short answer
On September 22, 2026, OpenAI introduced GPT-6 Sol and GPT-6 Luna, cheaper models trained with similar methods to GPT-6 Astra. API prices fell 50% versus GPT-5.6, to $2 and $10 per million tokens for Sol and $0.10 and $0.50 for Luna. They are in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu, with Luna also on Free and Go in the desktop app, but not yet in Chat.
GPT-6 Sol and GPT-6 Luna are OpenAI’s newest models, introduced on September 22, 2026, a little under three weeks after GPT-6 Astra. OpenAI says they bring Astra’s improvements to faster, cheaper models, and it has cut API prices in half. They are live in ChatGPT Work and Codex but not yet in ordinary chat. This report covers what OpenAI announced, using its own launch post, model docs and pricing page.
GPT-6 Sol and Luna at a glance
| GPT-6 Sol | GPT-6 Luna | |
|---|---|---|
| What OpenAI says it is for | Complex coding and agentic workflows, with stronger factual reliability and clearer communication than GPT-5.6 Sol | Focused, high-volume tasks such as summarisation, extraction and focused coding |
| API price (per million tokens) | $2 input, $10 output | $0.10 input, $0.50 output |
| Previous price | $4 and $20 (GPT-5.6 Sol) | $0.20 and $1.20 (GPT-5.6 Luna) |
| API name | gpt-6-sol | gpt-6-luna |
| In ChatGPT | Work and Codex on Plus, Pro, Business, Enterprise, Edu | Work and Codex on the same plans, plus Free and Go in the desktop app |
| In Chat | Not yet | Not yet |
What’s new
OpenAI says the GPT-6 models lead across the cost and intelligence curve, and that improvements in caching and inference let it pass savings on, cutting Sol and Luna API prices by 50% compared with GPT-5.6’s promotional pricing. It also says it brought Astra’s clearer communication style to both models, with less jargon, fewer odd phrases and slightly shorter answers.
OpenAI’s launch post makes several claims that are its own, run in its research environment or through the API, and it compares against Claude models that were current on the day, Claude Opus 5 and Fable 5.1, before Anthropic released Claude Opus 5.5:
- Business workflows: on AutomationBench, GPT-6 Sol at xhigh effort scores 33.2% at about $0.27 a task, above Claude Opus 5 at max effort at 26.9%.
- Coding: on DeepSWE v1.1, GPT-6 Sol at max effort scores 68.8%, within 1.1 points of Claude Fable 5’s best score, at roughly 80% lower cost per task, and GPT-6 Luna at max effort scores 66.6%.
- Factuality: OpenAI says GPT-6 Sol makes about half as many mistakes as its predecessor on its internal evaluation of conversations where users flagged errors.
Read those as a vendor’s own numbers.
Update: GPT-6.1 Sol
On September 29, 2026, OpenAI announced GPT-6.1 Sol, an upgrade to GPT-6 Sol at the same $2 and $10 API price, which it says nearly matches GPT-6 Astra on agentic coding, computer use and professional work. GPT-6 Luna is unchanged. See GPT-6.1 Sol for the details.
Who can use them in ChatGPT
OpenAI says GPT-6 Sol and Luna are available in ChatGPT Work and Codex starting September 22 for all Plus, Pro, Business, Enterprise and Edu users, and that Free and Go users can use GPT-6 Luna in the desktop app. They are not yet available in Chat, where the models are still the GPT-5.6 family and, on Pro plans, GPT-6 Pro. Rollout is gradual, so you may need to try again later. Our guide to ChatGPT models explained shows which model each plan uses where, and What is ChatGPT Work? covers the agent itself.
Codex usage estimates
OpenAI’s pricing page estimates local messages per five-hour period, and says they aren’t fixed limits:
| Model | Plus | Pro 5x | Pro 20x |
|---|---|---|---|
| GPT-6 Astra | 5-45 | 25-225 | 100-900 |
| GPT-6 Sol | 15-150 | 70-700 | 300-3,000 |
| GPT-6 Luna | 350-3,000 | 1,750-14,000 | 7,000-56,000 |
| GPT-5.6 Sol | 10-100 | 50-500 | 200-2,000 |
| GPT-5.6 Luna | 250-2,000 | 1,250-10,000 | 5,000-40,000 |
Luna supports reasoning efforts up to Max but not Ultra, the mode that uses subagents. See our guide to the best AI coding assistants for how Codex compares with Claude Code and others.
GPT-5.5 retires on October 14
OpenAI says GPT-5.5 retires from ChatGPT, Work and Codex on October 14, 2026, and advises choosing GPT-6 Sol on Plus, Pro, Business, Enterprise and Edu, or GPT-6 Luna in the desktop app on Free and Go, once available. It doesn’t affect the API. For plan details, see ChatGPT Pro explained and ChatGPT Go vs Plus, and for the assistant overall, the ChatGPT profile.
Frequently asked questions
What are GPT-6 Sol and GPT-6 Luna?
They are two models that extend the GPT-6 family beyond Astra. OpenAI says it trained them with similar methods, bringing Astra's advances in professional work, factuality, coding, computer use and alignment to faster, more affordable models. Sol suits complex coding and agentic workflows and Luna suits focused, repeatable tasks.
How much do GPT-6 Sol and Luna cost in the API?
GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens, down from $4 and $20 for GPT-5.6 Sol. GPT-6 Luna costs $0.10 and $0.50, down from $0.20 and $1.20 for GPT-5.6 Luna. OpenAI says both are 50% cheaper than the GPT-5.6 promotional pricing, though the listed Luna output price is a cut of about 58%.
Can I use GPT-6 Sol and Luna in ChatGPT?
In ChatGPT Work and Codex, yes, starting September 22 for Plus, Pro, Business, Enterprise and Edu users, with Free and Go users able to use GPT-6 Luna in the desktop app. OpenAI says they are not yet available in Chat, and that they roll out gradually.
What should I use instead of GPT-5.5, which retires on October 14?
OpenAI advises choosing GPT-6 Sol on Plus, Pro, Business, Enterprise and Edu plans, and GPT-6 Luna in the desktop app on Free and Go, when available. The API isn't affected by the retirement.
Do the API model names change?
The new models are gpt-6-sol and gpt-6-luna. OpenAI has also improved prompt caching for GPT-6, with 90% discounts on cached input-token reads.
How does usage compare with the older models in Codex?
OpenAI's estimates for Plus are 15 to 150 local messages per five hours on GPT-6 Sol against 10 to 100 on GPT-5.6 Sol, and 350 to 3,000 on GPT-6 Luna against 250 to 2,000 on GPT-5.6 Luna. OpenAI says these are estimates and not fixed limits.
Sources
Product facts are checked against the vendor's own announcements and documentation.
- Introducing GPT-6 Sol and Luna — OpenAIhttps://openai.com/index/introducing-gpt-6-sol-and-luna/
- Models — OpenAIhttps://learn.chatgpt.com/docs/models
- ChatGPT and Codex pricing — OpenAIhttps://learn.chatgpt.com/docs/pricing
