OpenAI introduces GPT-6.1 Terra for fast computer use

OpenAI's new mid-tier model reads screens and takes action across apps, and per OpenAI its API prices are half those of GPT-6.1 Sol.

GPT-6.1 Terra is out today as gpt-6.1-terra in the API. The OpenAI post calls it "a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing."

What changed#

@OpenAI announced it at 2:06 PM ET Monday. Per the post, standard API prices are "50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol": $1 input, $0.05 cached input and $5 output per million tokens. Per OpenAI, it scores 15 percentage points above GPT-5.6 Terra on Agents' Last Exam at medium effort, "at about one-third the cost".

Availability differs by page. The post says "available to all Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex". The release notes say it "is rolling out" and list Business too. The API changelog says "Released", while @OpenAIDevs says it "is also coming to the API and Codex".

Why it matters#

Our read: it's a cheaper default for screen-reading and form-filling agents. We flagged the name as a rumor on Oct 25.

How to use it#

Per OpenAI's release notes and GPT-6 guide:

In Work or Codex, "Select GPT-6.1 Terra from the model list."
In the API, set model to gpt-6.1-terra in a Responses request.
"Start with medium and increase it if the improvement in quality is worth the added time and tokens."
Move tool calls off Chat Completions: "migrate those requests to the Responses API".

Prompt (ours, untested), with an invoice image attached:

Read the attached invoice image. Return JSON with vendor name, invoice number, invoice date (YYYY-MM-DD), currency, total, and one object per line item with description, quantity and unit price. Use null for any field you can't read, and don't guess.

Good to know#

Per the release notes, "some Plus and Pro users may see GPT-6.1 Terra at Medium instead of GPT-6.1 Sol at Light in the default Power presets in Work".
"Terra does not support UAE inference residency", per the release notes.
Prompts over 272K input tokens cost "2x input and cache rates and 1.5x output for the full request", per the model page.
"For the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice", per the post.

Last checked: Oct 26, 2026, 3:58 to 4:00 PM ET. What changed: first published Oct 26; no updates since.

Sources#

Steps

In ChatGPT Work or Codex, select GPT-6.1 Terra from the model list once it reaches your account, per OpenAI's release notes.
In the API, set model to gpt-6.1-terra in a Responses API request, per OpenAI's GPT-6 guide.
Start with medium reasoning effort and increase it if the improvement in quality is worth the added time and tokens, per the guide.
If your application calls tools through Chat Completions, migrate those requests to the Responses API, per the guide.
When migrating from GPT-5.6 Terra, compare both models on representative images and workflows using the same reasoning effort, per the guide.