OpenAI released GPT-6 Sol and GPT-6 Luna on 22 September 2026, two lower-cost models in its GPT-6 family that sit below the flagship GPT-6 Astra. OpenAI says API prices are 50% lower than the GPT-5.6 versions’ promotional pricing: $2/$10 per million input/output tokens for Sol and $0.10/$0.50 for Luna. Both are live in the API, Codex and ChatGPT Work.
Last updated: September 2026
What did OpenAI announce?
According to OpenAI’s docs, Sol is built for complex coding and agentic workflows. OpenAI’s documentation describes Luna as its most efficient model for focused, high-volume tasks.
OpenAI says GPT-6 Astra remains its best model across the board, and that it built Sol and Luna using methods similar to Astra’s. The developer docs list the same technical envelope for both:
- Context: a 1,050,000-token context window and up to 128,000 output tokens per response.
- Inputs: text and images in, text out; no audio or video.
- Reasoning: adjustable effort from none to max.
- Knowledge cutoff: 20 April 2026 for Sol, 18 May 2026 for Luna.
- Tools: web search, file search, code interpreter, computer use and MCP connections through the Responses API.
OpenAI reports 68.8% for Sol on DeepSWE v1.1 and 60.5% on OSWorld 2.0, and compares these with Anthropic’s Claude Fable 5 and Claude Opus 5. These are vendor-reported results, not independent tests.
How much do GPT-6 Sol and Luna cost?
As of September 2026, OpenAI’s developer docs list these API prices per million tokens. Anthropic’s Claude Opus 5.5, launched the same day, is included for reference.
| Model | Input | Cached input | Output |
|---|---|---|---|
| GPT-6 Astra | $10.00 | $1.00 | $50.00 |
| GPT-6 Sol | $2.00 | $0.20 | $10.00 |
| GPT-6 Luna | $0.10 | $0.01 | $0.50 |
| Claude Opus 5.5 | $4.00 | $0.20 | $20.00 |
OpenAI credits caching and inference improvements for the 50% cut, which is measured against the GPT-5.6 models’ promotional pricing. Cached reads are 90% cheaper, and developers can change reasoning effort mid-conversation without invalidating the cache. OpenAI’s docs note that, for all three GPT-6 models, requests above 272,000 input tokens are billed at twice the input and cache rates and 1.5 times the output rate for the full request.
Who can use GPT-6 Sol and Luna in ChatGPT?
Both models are rolling out in ChatGPT Work, the agentic mode that plans and acts across tools, files and desktop apps, and in Codex for Plus, Pro, Business, Enterprise and Edu users. Free and Go users get Luna in the desktop app. OpenAI says neither model is in standard Chat mode yet. Developers call them as gpt-6-sol and gpt-6-luna.
How does GPT-6 Sol compare with Claude Opus 5.5?
TechCrunch reported that Anthropic released Opus 5.5 about 90 minutes before OpenAI’s announcement. Anthropic priced it 20% below Opus 5 per token, cut cache reads by 60%, and says Opus 5.5 costs 40% less to run than Opus 5 on typical workloads.
Per token, Sol’s input and output prices are half those of Opus 5.5, while cached input costs the same $0.20 per million on both. But OpenAI’s launch charts compare Sol with Opus 5 and Fable 5, not Opus 5.5, and The New Stack noted nobody had yet run the two head-to-head.
Why it matters for users and businesses
- Cheaper agents: coding and automation agents use many tokens, so half-price models lower the cost per task.
- Luna for volume work: Luna’s input price is 1% of Astra’s, which suits classification, extraction and ticket summarisation at scale.
For startups and agencies across the Gulf and wider MENA region running customer-service bots or Arabic-English document workflows on OpenAI’s API, lower unit costs can decide whether a pilot is affordable in production.
What happens next
OpenAI has not given a date for bringing Sol and Luna to standard Chat mode. Independent benchmarks comparing Sol and Opus 5.5 are next to watch.
Frequently asked questions
What is GPT-6 Sol?
GPT-6 Sol is a mid-tier OpenAI model released on 22 September 2026 for complex coding and agentic workflows. It costs $2 per million input tokens and $10 per million output tokens, has a 1,050,000-token context window, and is available in the API, Codex and ChatGPT Work for paid plans.
How much does GPT-6 Luna cost?
As of September 2026, GPT-6 Luna costs $0.10 per million input tokens, $0.01 per million cached input tokens and $0.50 per million output tokens. According to OpenAI’s documentation, requests above 272,000 input tokens are billed at twice the input and cache rates and 1.5 times the output rate for the full request.
Can free ChatGPT users use GPT-6 Luna?
Yes, but only in the ChatGPT desktop app. OpenAI says Free and Go users can access Luna there, while Sol and Luna in ChatGPT Work and Codex are for Plus, Pro, Business, Enterprise and Edu users. Neither model is in standard Chat mode yet.
Is GPT-6 Sol better than GPT-6 Astra?
No. OpenAI says GPT-6 Astra remains its best model across the board, and its docs describe Astra as built for the hardest end-to-end work. Sol is built for complex coding and agentic workflows at a fifth of Astra’s per-token price: $2/$10 per million input/output tokens versus $10/$50 for Astra.
Is GPT-6 Sol cheaper than Claude Opus 5.5?
Per token, yes. Sol is priced at $2/$10 per million input/output tokens versus $4/$20 for Claude Opus 5.5; cached input is $0.20 per million on both. Real costs also depend on how many tokens each model uses per task and on caching, so test both on your own workload.
Sources
- OpenAI: Introducing GPT-6 Sol and Luna
- OpenAI API docs: GPT-6 Sol
- OpenAI API docs: GPT-6 Luna
- OpenAI API docs: GPT-6 Astra
- OpenAI: ChatGPT Work
- Anthropic: Claude Opus 5.5
- TechCrunch: OpenAI launches GPT-6 Sol and Luna
- The New Stack: OpenAI releases GPT-6 Sol and Luna
- MacRumors: OpenAI’s new GPT-6 Sol and Luna models
