TL;DR

  • GPT-6 Sol is OpenAI's model for complex coding, reasoning, and agentic workflows, released September 22, 2026 with GPT-6 Luna.
  • It lists at $2 per million input tokens and $10 per million output tokens, 50% below GPT-5.6 promotional pricing.
  • It has a 1.05M token context window and 128K token maximum output.
  • Prompts over 272K tokens are billed at doubled input rates and 1.5x output.
  • Batch and Flex processing cost 50% of standard rates, and Fast mode costs 2x.
  • It is available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, and in the API.

OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, 2026, the same day Anthropic shipped Claude Opus 5.5. Sol is the middle of the GPT-6 family: OpenAI’s model docs describe it as designed for complex coding and agentic workflows, and the announcement says Sol and Luna carry much of GPT-6 Astra’s strength into faster, more affordable models built for work at scale.

What is GPT-6 Sol?

Sol is a general-purpose GPT-6 model with the family’s full toolkit. OpenAI’s docs list a 1.05M token context window, a 128K token maximum output, vision, reasoning with adjustable effort, and the standard tool surface: function calling, web search, file search, and computer use. It is served through the Responses API and Chat Completions.

The lineage matters for expectations. OpenAI says Sol and Luna build on the advances behind GPT-6 Astra, including Astra’s alignment work, and that both show improvements over their GPT-5.6 counterparts. In practice, Sol is the model you reach for when a GPT-5.6-era workflow needs more capability without jumping to Astra pricing.

Availability, per the announcement: GPT-6 Sol and Luna rolled out in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, and both are in the API. Free and Go users can try GPT-6 Luna in the desktop app.

How much does GPT-6 Sol cost?

Standard rates, per 1M tokens:

Item Price
Input $2
Cached input $0.20
Cache write $2.50
Output $10

OpenAI also cut the whole family’s floor: the announcement says Sol and Luna API prices are 50% below GPT-5.6 promotional pricing.

Three adjustments change the real bill. First, long prompts get a step rate: for prompts over 272K input tokens, input and cache rates double and output is billed at 1.5x. Second, Batch and Flex processing cost 50% of standard, so async workloads such as overnight processing or bulk enrichment are effectively half price. Third, Fast mode costs 2x for lower latency. If your feature sends a lot of context, the 272K step is the one to model before you commit, since a prompt cache can also turn repeat context into the $0.20 cached rate.

Where does Sol sit in the GPT-6 family?

Model List price, per 1M tokens Built for
GPT-6 Luna $0.10 input, $0.50 output high-volume, focused tasks such as extraction, classification, and simple coding
GPT-6 Sol $2 input, $10 output complex coding, reasoning, and agentic workflows
GPT-6 Astra $10 input, $50 output the hardest end-to-end reasoning and professional work

All three share the 1.05M context window, 128K maximum output, and the tool surface, per OpenAI’s models overview. The spread is wide enough that routing matters: at list price, Sol costs 20 times Luna’s input rate, and Astra costs 5 times Sol’s. A product that routes classification traffic to Luna and hard generation work to Sol keeps most of Astra-class quality for a fraction of Astra-class spend. The deeper comparison lives in GPT-6 Sol vs GPT-6 Luna.

What should a product team weigh before switching?

Cost of goods first. A feature that sends 10K input tokens and generates 2K output tokens per interaction costs about $0.04 at Sol’s list price. Multiply by real interaction volume before assuming the switch is cheap, and remember that an AI feature inside a native app carries costs beyond tokens: if the feature unlocks digital functionality, store billing rules apply regardless of which model powers it. App monetization covers that boundary.

Then migration surface. Sol is API-compatible with the GPT-5.x integration patterns through Responses and Chat Completions, and reasoning effort is a parameter rather than a redesign, so most teams can move gradually: route one workflow, watch quality and cost, then widen.

If you are also weighing the same-day Anthropic release, Opus 5.5 vs GPT-6 Sol puts the two side by side.

Sources

FAQ

What is GPT-6 Sol?
GPT-6 Sol is one of OpenAI's GPT-6 models, released September 22, 2026. OpenAI positions it for complex coding, reasoning, and agentic workflows, with much of GPT-6 Astra's capability at a lower price.
How much does GPT-6 Sol cost?
The API lists $2 per million input tokens and $10 per million output tokens, with cached input at $0.20. Prompts over 272K tokens are billed at doubled input rates and 1.5x output rates. Batch and Flex processing cost 50% of standard, and Fast mode costs 2x.
What is the difference between GPT-6 Sol and GPT-6 Luna?
Sol is built for complex coding and agentic workflows. Luna is built for high-volume, focused tasks such as extraction and classification, and lists at $0.10 per million input tokens and $0.50 per million output tokens.
Does GPT-6 Sol have a one million token context window?
Slightly more. OpenAI's model docs list a 1,050,000 token context window and a 128,000 token maximum output for GPT-6 Sol.