GPT-6.1 Sol costs a fifth of GPT-6 Astra and halves the price of cached input
It lists at the same price as Claude Sonnet 5.5, but its cached input now costs half as much.
OpenAI released GPT-6.1 Sol on September 29, 2026, and priced it at $2 per million input tokens and $10 per million output. That's a fifth of what GPT-6 Astra costs, and OpenAI's own model page sells it as "Near-Astra performance for complex work at a lower cost."
The list price didn't move from GPT-6 Sol. The cache did. Cached input now costs $0.10 per million tokens, half the old Sol's $0.20, and that's the number Altman led with.
- Input
- $2.00 per 1M tokens
- Output
- $10.00 per 1M tokens
- Cached input
- $0.10 per 1M tokens
- Context window
- 1,050,000
- Knowledge cutoff
- Apr 30, 2026
Where it sits against Astra
OpenAI's page says 6.1 Sol is for complex coding, computer use and professional work, and tells developers to compare it with Astra on their own tasks to judge the tradeoff between quality and cost. That's a fair warning.

On paper the two are close. Both take 1,050,000 tokens of context and write up to 128,000 tokens back, and both have an April 30, 2026 knowledge cutoff. Astra is $10 in and $50 out.
Some settings are gone, too. 6.1 Sol drops the none and minimal reasoning levels that GPT-6 Sol offered, so code that ran the old Sol at none has to move to low. Tool calls need the Responses API.
OpenAI's own numbers against Opus 5.5
Update, October 1, 2026: OpenAI's launch post, which wouldn't load for us on launch day, sets 6.1 Sol against Anthropic's Claude Opus 5.5 on three tests. On AutomationBench, which checks whether an agent finishes multi-step business workflows, OpenAI puts Sol 2.2 percentage points ahead of Opus 5.5 at medium reasoning effort, for about a third of the cost. On GDP.pdf (professional questions asked about dense PDFs full of tables and fine print), it says Sol beats Opus 5.5 run with fallbacks, at under half the cost per task.
The third comparison is about money. On Terminal-Bench Science 0.1 at maximum effort, Sol averaged $5.47 a task, against $23.21 for Opus 5.5 and $23.80 for Astra. Astra still has the top score there, 68.1%, and OpenAI tells people to use it for the hardest research work. The post also says its competitor figures were taken from publicly available reports rather than run by OpenAI.
The comparison that changed
Anthropic launched Claude Sonnet 5.5 the day before, at the same $2 and $10. Its cache reads cost $0.20 per million tokens.
Sonnet 5.5 is priced the same as Sonnet 5 at $2 per million input tokens, $10 per million output tokens, and $0.20 per million tokens for cache reads
So GPT-6 Sol and Sonnet 5.5 matched line for line. Now OpenAI's cached input is half Anthropic's. That matters most to agents and coding tools, which send the same long prompt over and over. Anthropic's counter is tokens: it says Sonnet 5.5 typically needs far fewer of them to do the same work, which is why our Sonnet 5.5 brief led with cost per task.
What Altman said
He posted a minute or two before his dots announcement, and kept it to the price.
6.1 Sol at 1/5th of the price of Astra, and 95% cache read discount! Extremely capable model.
You call it as gpt-6.1-sol. OpenAI says it supports US and EU data residency, though fast mode isn't available with the EU option.
More on OpenAI
- ChatGPT Wallet lets OpenAI's dots buy things for you, but you approve each purchaseNovember 14, 2026
- Helen Toner would vote to remove Altman again, saying the future can't hinge on one CEONovember 14, 2026
- Albanese says Australia will create an AI safety regulator that can impose significant penaltiesNovember 10, 2026
- OpenAI and Anthropic each say an internal AI model disproved the Exponential Time HypothesisNovember 7, 2026
