OpenAI says GPT-6.1 Terra can fill out forms and edit videos for paid users
OpenAI says its new GPT-6.1 Terra model can look at a screen and carry out computer tasks. It is on paid ChatGPT Work and Codex plans and for developers, and the test figures are OpenAI's own.
A new model can look at what is on a screen and carry out everyday computer work, from filling out forms to editing videos, its maker OpenAI says. In its announcement, the company calls GPT-6.1 Terra "a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing".
One example OpenAI gives is reading incoming invoices and entering their details into accounting software, work it says the model can take on.
It is available on paid plans (Plus, Pro, Enterprise and Edu) in ChatGPT Work and Codex. Developers can use it as gpt-6.1-terra through the API, the route for connecting their own software to the model.
An outside analysis by CellCog notes that the announcement does not name the Free or Business ChatGPT plans. It reports a release date of 26 October 2026.
Prices are counted in tokens, the small pieces of text a model reads and writes. Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens (text the model has already been given and reuses) and $5 per million output tokens. OpenAI says these prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol.
The performance figures are OpenAI's own runs and cost estimates, CellCog points out, and no outside lab had published results when it wrote.
OpenAI measured computer use on Agents' Last Exam, a set of professional tasks that includes turning architectural drawings into a 3D model and reviewing an animated film shot by shot to make a time-coded log for a restoration team.
In OpenAI's own tests on that set, the model at medium reasoning effort scored 15 percentage points higher than GPT-5.6 Terra at about one-third of the cost. Reasoning effort is a setting for how much effort the model puts into a task.
OpenAI names a limit: for the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice. CellCog reports that the model takes text and images in and gives text out, and that audio and video are not supported as inputs.
On safety, OpenAI says its prompt-injection tests showed strong resistance to malicious instructions designed to pull the model away from the user's task. It says the model keeps the protections developed for GPT-6 Astra, and meets what its Preparedness Framework, a system for rating risky abilities, calls the Critical threshold in cybersecurity.
OpenAI's system card reports that the model scored worse than GPT-5.6 Terra on tests of hate and extremism that combine text and images. OpenAI says a manual review found performance in line with prior launches.
Sources
- Introducing gpt 6 1 terra · openai.com
- Gpt 6 1 terra · cellcog.ai
- Gpt 6 1 terra · deploymentsafety.openai.com