What does GPT-6.1 Sol cost?

OpenAI's announcement of September 29, 2026 lists GPT-6.1 Sol at $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens. OpenAI's Enterprise rate card lists GPT-6 Astra at $10, $1 and $50, so Sol is one fifth of Astra's input and output price.

Does GPT-6.1 Sol lower a ChatGPT Business bill?

No. ChatGPT Business is a flat seat, $20 a month on annual billing for a standard seat and $100 for a premium seat on OpenAI's pricing page on September 30, 2026. A cheaper model does not change the seat price.

Does GPT-6.1 Sol lower a ChatGPT Enterprise bill?

It can, on token-based Enterprise agreements. OpenAI's rate card charges ChatGPT Work and Codex activity by the model used, so work that moves from GPT-6 Astra to GPT-6.1 Sol is billed at a fifth of the input and output rate. Enterprise agreements on credits follow a separate rate card.

Is GPT-6.1 Sol available in regular ChatGPT chat?

Not yet. OpenAI's announcement says GPT-6.1 Sol is available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users, and through the API. It is not yet available in Chat, so everyday chat usage is not affected.

GPT-6.1 Sol is a fifth of Astra's price. Who saves?

On September 29, 2026, OpenAI released GPT-6.1 Sol and priced it at a fifth of its top model, GPT-6 Astra. For a company, that only lowers the bill where usage is charged by the token, and only for work that moves off Astra. A ChatGPT Business seat costs the same as it did last week.

What did OpenAI change with GPT-6.1 Sol?

The figures below are from OpenAI’s announcement of September 29, 2026 and its Enterprise rate card and pricing page as checked on September 30, 2026. Check the source before you budget on them.

GPT-6.1 Sol is $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens.
GPT-6 Astra is $10, $1 and $50 on the rate card. Sol is one fifth of Astra’s input and output price and one tenth of its cached input price.
The list price is the same as GPT-6 Sol, released on September 22, 2026. Only cached input fell, from $0.20 to $0.10.
OpenAI says the new model nearly matches Astra on agentic coding, computer use and professional work. That is OpenAI’s own testing.
It is available to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, and through the API. It is not yet available in Chat.

Anthropic’s announcement of September 28, 2026 lists Claude Sonnet 5.5 at $2 and $10 as well. The two labs now charge the same list price for their everyday work model.

A company with both on hand can pick on results. Harriet runs Claude, GPT, Gemini and open models through one router, GPT-6.1 Sol and Astra included, so moving a team from one to the other is a setting on the team profile.

Does a ChatGPT Business bill go down?

No. OpenAI’s pricing page on September 30, 2026 lists a standard Business seat at $20 a month on annual billing and $25 on monthly billing, and a premium seat at $100 and $125. The page describes the premium seat as five times the usage of a standard seat.

Those are flat fees. A cheaper model gives each seat a more capable tool in ChatGPT Work and Codex. It does not change the invoice. The announcement does not mention any change to Business usage limits.

A company that reaches OpenAI’s models with its own API key pays the token price, which is where this release lands. That is how Harriet bills when you bring your own key, with no model charge from Harriet on top.

Does a ChatGPT Enterprise bill go down?

It depends on how your agreement is billed. OpenAI’s pricing page says Enterprise is custom priced, with credit-based and token-based billing both available.

On a token-based agreement, the rate card charges ChatGPT Work and Codex activity by the model used and the tokens consumed. A task that moves from Astra to GPT-6.1 Sol is billed at a fifth of the input and output rate. If your teams were already on GPT-6 Sol, only the cached input line changes.

On a credit-based agreement, OpenAI publishes a separate rate card. Check yours before you assume the same ratio.

Where does the saving not reach?

Chat, fast modes and long context. All three are priced on OpenAI’s rate card.

GPT-6.1 Sol is not in Chat. The rate card’s Chat table lists the previous generation, GPT-5.6 Sol, at $4 and $20, and Astra at $10 and $50. A company whose people mostly chat sees no change from this release.

Fast mode is billed at twice the standard token rate. OpenAI’s DevDay recap of September 29, 2026 says an Ultrafast tier for GPT-6.1 Sol is coming soon, and the rate card already prices Astra Ultrafast at six times standard. At twice the rate, GPT-6.1 Sol in Fast mode costs $4 and $20.

Above 272,000 input tokens, the rate card doubles the input rate and charges one and a half times the output rate.

What should a company do this week?

Find out how much of your usage is Chat and how much is ChatGPT Work and Codex. Only the second can move to GPT-6.1 Sol today.
On Enterprise, decide which teams keep Astra. Astra is also the model behind OpenAI’s dots, where workspace model controls do not apply. The rate card says workspace admins can enable or disable access to each model. Use that, and decide who gets Fast mode.
Test before you switch. Take ten real tasks, run them on Astra and on GPT-6.1 Sol, and compare the output and the token count. OpenAI’s benchmarks are where that test starts. Harriet’s eval mode runs that comparison for you, with output, tokens, cost and time side by side.

Anthropic and OpenAI shipped a cheaper everyday model a day apart. In both cases the saving goes to companies that pay for usage and choose which work goes to which model.

On Harriet that choice sits with the admin, who sets the default model and a budget for each team. The router adds new models as they ship.

For what routing and caps do to an AI bill, start with cost control. For how OpenAI prices its top plan, see ChatGPT Enterprise pricing. If you are choosing between the two vendors, read ChatGPT Enterprise vs Claude Enterprise.