The story in brief
- 1OpenAI released GPT-6.1 Terra in late October 2026, a fast model for computer use, visual reasoning and writing.
- 2API pricing is $1 per million input tokens, $0.05 cached and $5 output, half the price of GPT-6.1 Sol and of the previous Terra.
- 3It is available in the API as gpt-6.1-terra, and in ChatGPT Work and Codex on paid plans.
- 4OpenAI says it stays within task scope and reports blockers more reliably, and that it resists prompt injection in its tests.
- 5OpenAI still recommends its larger models for the hardest coding work, and its benchmark claims have not been independently reproduced.
OpenAI has released GPT-6.1 Terra, a fast model built for computer use, visual reasoning and writing. In the API it costs $1 per million input tokens and $5 per million output tokens, half the price of GPT-6.1 Sol. For businesses, cheaper computer use makes AI agents that work through old portals and desktop software more affordable, as long as they are given the right checks.
What OpenAI released
- Model: GPT-6.1 Terra, available in the API as `gpt-6.1-terra`. It replaces GPT-5.6 Terra.
- Built for: computer use, meaning reading what is on a screen and operating applications, plus visual reasoning and writing.
- Price: $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. That is half the price of GPT-6.1 Sol, and half of the previous Terra. Reports of OpenAI's pricing add that Batch and Flex processing cost 50% less again.
- Where: the OpenAI API, and ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users.
What OpenAI says about safety and limits
OpenAI says Terra is more likely than the previous version to stay within the task it was given and to report when it is blocked, and that it showed strong resistance to prompt injection in OpenAI's own tests. It also says its larger models are still the better choice for the most complex coding work. The benchmark results OpenAI published have not been independently reproduced yet, so treat them as claims.
Why computer use matters to businesses
Many business systems have no API: an old desktop ERP, a supplier portal, a government website. Until now, automating them meant a fragile script or a person copying data by hand. A computer-use agent reads the screen and clicks and types like a person. Cheaper models make that affordable for everyday tasks such as:
- pulling reports from a portal every morning,
- keying invoices or orders into a system with no import,
- checking statuses across several websites.
It is still slower than an API and can still misread a screen, so the same rules apply: check anything with a right answer in code, have a person approve anything that goes to a customer or moves money, and log every action. Our guide to AI agent security basics covers the sign-in rules.
What it means for businesses
Cheaper models change which tasks are worth automating, but the model is only one part of an agent. The work is in connecting it to your systems, choosing what it may do and checking what it did. Our AI agent development team builds agents like this, picks the model per task on cost and accuracy, and tests a new model on your real data before switching. For examples of what we have built, see what an AI automation agency does.
Sources
Frequently asked questions
A fast, lower-cost OpenAI model built for computer use, visual reasoning and writing. It can read what is on a screen and operate applications to complete tasks, and it replaces GPT-5.6 Terra.
In the API, $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens, which is half the price of GPT-6.1 Sol.
That is what computer use is for: the agent reads the screen and clicks and types like a person, so it can work with old portals and desktop software. It is slower and less reliable than an API, so it still needs checks and approvals.
Test it first on your own tasks. A cheaper model only saves money if it gets your tasks right, so compare cost and accuracy per task on real data before switching.
Related services
Get new articles by email
How we build automation and software, with the lessons learned. One short email when a new article goes up. No spam.
- #OpenAI
- #GPT-6.1 Terra
- #AI Agents
- #Computer Use
- #AI Pricing
