<MCP.Playground/>
online

GPT-6 Sol

Runs MCP tools

OpenAI’s mid-tier GPT-6, released 22 September 2026 and tuned for recurring coding and agent work. It sits below the flagship Astra and above the fast Luna tier.

Vendor
OpenAI
Released
22 September 2026
Context
1.05M tokens
Input
Text, image, file

Paste a server URL, pick a model, and watch it call your tools in a real conversation. No install.

No server of your own? Leave it blank and use one of the public mock servers.

MCP Evals

Can GPT-6 Sol actually complete tasks with your server?

One chat shows the tool calls work. An eval writes real tasks from your tool schemas, has GPT-6 Sol drive each one, and reports where it got the wrong answer even though every call succeeded. Pick GPT-6 Sol as the driver.

Run evals →See a sample

What GPT-6 Sol is

GPT-6 Sol is the high end of OpenAI’s GPT-6 series. Astra is the flagship above it and Luna the fast tier below. It launched the same day as Claude Opus 5.5, with a 1.05M-token window. Sol Pro is the same model served with reasoning mode set to pro. It spends more reasoning on each answer.

What OpenAI says

  • AutomationBench: 33.2% of tasks.
  • DeepSWE v1.1: 68.8%.
  • About half as many mistakes as its predecessor, by OpenAI’s account.

These are the vendor’s own claims, not measurements of ours. Run the model against your own server to find out whether they hold for your tools.

How it reaches MCP

  • The OpenAI Responses API, which can call remote MCP servers as a tool type. Model id gpt-6-sol.
  • ChatGPT, where Sol is available on the Work, Pro, Business, Enterprise and Edu plans and MCP servers connect as connectors
  • Any MCP client over OpenRouter as openai/gpt-6-sol or openai/gpt-6-sol-pro, which is what runs here

What to watch for

  • Sol Pro reasons for longer, so each run takes more time and output. Test the one you plan to ship. The two can pick tools differently on the same prompt.
  • OpenRouter lists seed but not temperature for the GPT-6 models. A client that lowers temperature to make tool calls repeatable gets no effect here. Use seed instead.
  • It supports structured_outputs, so strict JSON-schema arguments are available. If your tool schema uses keywords that strict mode rejects, the failure shows up as a schema error, not a bad call.

Give Sol and Sol Pro the same multi-step prompt against the complex-schema mock, for example create a profile and then configure_workflow for it. The Pro run shows whether extra reasoning changes which tools get called, not just how the answer reads. All the mock servers →

Variants available here

ModelModel ID
GPT-6 Solopenai/gpt-6-sol
GPT-6 Sol Proopenai/gpt-6-sol-pro

Go deeper

Frequently asked questions

What is the difference between GPT-6 Sol and Sol Pro?
Same model, different reasoning mode. Sol Pro runs with reasoning mode set to pro for harder tasks. Pro produces more reasoning per answer, so runs are slower and use more tokens.
GPT-6 Sol or GPT-6 Astra?
Astra is OpenAI’s flagship for the hardest end-to-end work. OpenAI tunes Sol for recurring coding and agent work, which is what most MCP workloads are, so test Sol first and move up only if it misses on your tools.
Does GPT-6 Sol work with MCP servers?
Yes. OpenAI’s Responses API can call remote MCP servers, ChatGPT supports them as connectors on paid plans, and any MCP client can drive it through function calling. You can run it against your server here without an OpenAI key.
When was GPT-6 Sol released?
22 September 2026, alongside GPT-6 Luna and on the same day Anthropic released Claude Opus 5.5.

Sources

GPT-6 Sol for MCP: Specs & Tool Calling — Test Free | MCP Playground