SNACK: 3-line summary
- OpenAI has launched GPT-6.1 Sol for Codex, ChatGPT Work and the API.
- Plus, Pro, Business, Enterprise and Edu users can access it in Work and Codex; developers can use the gpt-6.1-sol API model.
- Regular ChatGPT chat does not support it yet, and Ultrafast is still to come.
GPT-6.1 Sol is available now for coding and work through Codex, ChatGPT Work and the API. If you’re looking for it in an ordinary ChatGPT conversation, it isn’t there yet.
Snackgirls react
AIKO: The halved cached-input rate catches my eye. I’d want to see how much context a real workflow can reuse before estimating its costs.
Red: I’d like to give it a coding task in Codex and see how it handles the messy parts. I won’t go hunting for it in regular ChatGPT chat yet.
Who can use it now
OpenAI says Plus, Pro, Business, Enterprise and Edu users can access GPT-6.1 Sol in ChatGPT Work and Codex. Developers can select the API model gpt-6.1-sol. OpenAI says Ultrafast is coming in the next few days, but it is not available yet.
What changes from GPT-6 Sol
In OpenAI’s DeepSWE 1.1 evaluation, GPT-6.1 Sol scored 6.4 percentage points above GPT-6 Sol’s best result while using lower reasoning effort and costing less. OpenAI also says it matched GPT-6 Astra on that test at roughly one-fifth the cost per task. These are OpenAI’s benchmark results, not a guarantee of the same improvement on every project.
GPT-6.1 Sol’s standard API prices are $2 per million input tokens and $10 per million output tokens. Cached input costs $0.10 per million tokens, 50% less than GPT-6 Sol’s cached-input price. OpenAI separately describes GPT-6.1 Sol’s standard input and output token prices as one-fifth of Astra’s; that comparison does not mean every task costs one-fifth as much.
Sources and date: 2026-09-30 KST