GPT-6.1 Terra: Computer Use for Office Tasks, Pricing, and Alternatives (2026)
GPT-6.1 Terra is OpenAI's new fast, lower-cost model for computer use and everyday writing. OpenAI released it on October 26, 2026, in ChatGPT Work, Codex, and the API. Its standard API price is $1 per million input tokens and $5 per million output tokens, half the price of GPT-6.1 Sol.
This guide covers what GPT-6.1 Terra does, which office tasks OpenAI built it for, and what it costs. It closes with the main alternatives and where each one fits. Every detail comes from the vendors' own pages, read on October 27, 2026.
What OpenAI announced on October 26
OpenAI calls it "a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing." It is the successor to GPT-5.6 Terra in the same model family.
The release has five main parts:
- Computer use at a lower cost. OpenAI says the model is "Delivering fast and capable computer use at scale."
- Office tasks as the target. The post says it excels at work "from drafting proposals and interpreting technical drawings to filling out forms and editing videos."
- Half-price API. Its "standard API prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol."
- More work per plan in ChatGPT. In ChatGPT Work and Codex, "the same usage allowance covers more tasks than with GPT-5.6 Terra."
- Fewer factual errors. OpenAI reports large drops in its internal error-rate test, covered below.
The ChatGPT release notes list the rollout under October 26, 2026, a Monday. OpenRouter also lists the model with a release date of October 26.
Computer use for office tasks: what GPT-6.1 Terra is built to do
Computer use means the model operates apps on screen. It reads what is displayed, then clicks, types, and moves between windows to finish a task.
OpenAI gives one office example in detail. The low API price makes it "practical for developers to automate high-volume work, like reading incoming invoices and entering their details into accounting software." That is a classic back-office job: read a document, pull the fields, and key them into a system.
The other examples in the post point the same way:
- Drafting proposals. This leans on the writing gains OpenAI describes.
- Interpreting technical drawings. This leans on visual reasoning.
- Filling out forms. This combines reading, typing, and checking fields.
- Editing videos. This shows the model can work inside creative desktop apps.
OpenAI measured these skills on Agents' Last Exam, "a benchmark of professional tasks." At medium reasoning effort, OpenAI says GPT-6.1 Terra "scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost."
OpenAI also compares it with larger models. It says the model "performs on par with GPT-6 Astra at only one-tenth its price per token." At max effort, OpenAI reports "a ~9x lower cost per task" than Claude Opus 5.5. These are OpenAI's own results on its chosen tests, so treat them as a starting point for your own trial.
Speed is the other claim. The post says Terra completes tasks "over three times faster than both GPT-6 Astra and GPT-6.1 Sol." For a person watching an agent work through a form, speed matters as much as accuracy.
Writing, accuracy, and coding changes
Computer use is the headline, but OpenAI lists three other changes.
Writing. The launch post promises "clearer, more natural writing." It also says the model excels at "communication and computer-based tasks that fill the workday." Proposals, emails, and summaries fall into that group.
Factual accuracy. OpenAI reports results from an internal test of "error-inducing prompts." At low reasoning effort, the error rate "fell from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra." At maximum effort, it "fell by 18 percentage points to 4.3%."
Coding. On Terminal-Bench 4.0, GPT-6.1 Terra "solves 50.3% of tasks" at maximum effort. That is "more than double GPT-5.6 Terra's published score of 21.5%." OpenAI is also clear about the limits here. It says "For the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice."
Safety rating. The system card says OpenAI is "treating GPT-6.1 Terra as Critical capability in Cybersecurity, High capability in the Biological and Chemical domain." It uses the same safeguards as GPT-6.1 Sol and GPT-6 Astra.
How to access GPT-6.1 Terra
There are three ways to use the model.
ChatGPT Work. The release notes say GPT-6.1 Terra "is rolling out in ChatGPT Work and Codex." The rollout covers "Plus, Pro, Business, Enterprise, and Edu plans." You pick it from the model list once it reaches your account.
Codex. The same rollout covers Codex, OpenAI's coding agent. OpenAI still points the hardest coding work to Sol and Astra.
The OpenAI API. Developers can call it as gpt-6.1-terra. OpenRouter also lists it for teams that route requests across several providers.
Three details from the release notes are worth knowing:
- "Your plan's overall usage limits are unchanged."
- In Work's default Power presets, some Plus and Pro users "may see GPT-6.1 Terra at Medium instead of GPT-6.1 Sol at Light."
- "Terra does not support UAE inference residency."
Note the small gap between sources. The launch post names Plus, Pro, Enterprise, and Edu. The release notes add Business. Check your own workspace settings if you are on a Business plan.
GPT-6.1 Terra pricing
OpenAI's API pricing page lists GPT-6.1 Terra next to its other current models. The page says the prices reflect "standard processing rates for context lengths under 272K."
| Model | Input, per 1M tokens | Cached input, per 1M tokens | Output, per 1M tokens |
|---|---|---|---|
| GPT-6.1 Terra | $1.00 | $0.05 | $5.00 |
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 |
| GPT-6 Astra | $10.00 | $1.00 | $50.00 |
| GPT-6 Luna | $0.10 | $0.01 | $0.50 |
Terra sits between Luna and Sol. Luna is OpenAI's "most efficient model for focused, high-volume tasks." Sol is "Built to power complex coding and agentic workflows at scale."
In ChatGPT, there is no separate price. The model comes with the plans listed above, and it draws on your plan's existing usage allowance. OpenAI's pitch is that the same allowance now covers more tasks.
Here is a rough way to think about API cost. An invoice-entry run might read a page image and write a short structured record. The output side is usually small, so the input rate drives most of the bill. Test with a batch of your own documents before you estimate a monthly cost.
Which tasks suit GPT-6.1 Terra
Match the model to the job, not to the leaderboard.
Good fits. High-volume, repeatable tasks with clear steps. Examples include invoice entry, form filling, and checking fields across two systems. Short drafting work, such as proposals and follow-up emails, also fits.
Weaker fits. OpenAI itself points the most complex coding to GPT-6.1 Sol and GPT-6 Astra. Long research that needs deep reasoning may also justify a larger model.
Work that needs a finished output. Many office tasks end in a report, a table, or a slide deck. A model that clicks through apps is only part of that job. You still need the analysis, the checks, and a format your readers expect.
Before rolling it out, run a small trial:
- Pick ten real tasks your team repeats each week.
- Run each one with GPT-6.1 Terra and with the model you use today.
- Count the fixes you had to make, not only the time saved.
- Decide per task, not for the whole team.
What to watch next
Three things are worth tracking after launch.
Wider availability. The rollout reaches accounts gradually. Watch the ChatGPT release notes for Business workspace details and any new regions.
Real-world error rates. OpenAI's accuracy numbers come from an internal test. Your own documents and forms will show how the model handles messy inputs, such as scanned invoices.
Pricing changes. OpenAI's pricing page also lists Batch, Flex, and data residency options. If you run large jobs overnight, compare those modes against the standard rate.
GPT-6.1 Terra alternatives
Several other options cover similar office and analysis work. The table uses each vendor's own pricing page, read on October 27, 2026.
| Option | What it offers | Price, per the vendor |
|---|---|---|
| GPT-6.1 Sol | OpenAI's model for complex coding and agentic work | API: $2 input and $10 output per million tokens |
| Claude Opus 5.5 | Anthropic's first model in its Claude 5.5 family | API: $4 input and $20 output per million tokens |
| Claude Sonnet 5.5 | Anthropic's faster, lower-cost model | API: $2 input and $10 output per million tokens |
| Kimi K3.1 Pro | Moonshot's open-weight model with a 1M context window | API: $2 input and $10 output per million tokens |
| DeepSeek V4.1 Flash | A low-cost open model | API: $0.3 input and $1.2 output per million tokens at peak; off-peak is half |
| Powerdrill Bloom | An AI workspace that turns files into analysis, reports, and slides | Free $0.00; Pro $13.27/mo |
GPT-6.1 Sol is the natural step up inside OpenAI's own lineup. Choose it when coding depth matters more than speed.
Claude Opus 5.5 is the model OpenAI used as its outside comparison. Our Claude Opus 5.5 guide covers its launch and pricing. Claude Sonnet 5.5 costs the same per token as GPT-6.1 Sol.
Kimi K3.1 Pro launched three days earlier, with open weights due by November 4. Our Kimi K3.1 Pro guide covers its finance focus.
DeepSeek V4.1 Flash is the lowest-cost API option here. DeepSeek's pricing page says off-peak rates "are half of the peak rates."
For context on OpenAI's top model, see our GPT-6 Astra guide.
Where Powerdrill Bloom fits
GPT-6.1 Terra is a model. Powerdrill Bloom is a workspace that delivers the finished output many office tasks end in.
Take OpenAI's invoice example. Terra reads invoices and keys the fields into accounting software. A finance lead then needs the totals, the trends, and a summary for the month. Powerdrill Bloom covers that second part. You upload the files and ask in natural language. Its homepage says "Every number comes back with the page, the row and the figure behind it."
Three pages match office work that starts where Terra stops:
- The PDF to Excel converter is built to "Turn PDF into Excel in Seconds."
- The Excel to PPT tool turns a spreadsheet into "presentation-ready slides automatically."
- The AI Financial Report Generator builds a "Monthly close report PDF with narrative, ratios and call-outs."
For repeat work, its Scheduled Agent reruns an analysis on a schedule. The changelog says "Every run finishes on schedule and delivers the results to your workspace."
On price, the pricing page lists a Free plan at $0.00 and Pro at $13.27/mo. Pro adds monthly credits and 20 scheduled tasks. To compare it with a model you are testing, try Powerdrill Bloom with one month of invoices you already know well.
Frequently asked questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is an OpenAI model released on October 26, 2026. OpenAI describes it as fast and efficient, with visual reasoning, precise computer use, and clearer writing. It is available in ChatGPT Work, Codex, and the OpenAI API.
How much does GPT-6.1 Terra cost?
On the API, GPT-6.1 Terra costs $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. In ChatGPT, it is included in Plus, Pro, Business, Enterprise, and Edu plans.
Is GPT-6.1 Terra better than GPT-6.1 Sol?
It depends on the task. OpenAI says GPT-6.1 Terra is faster and cheaper, and strong at computer use and writing. For the most complex coding tasks, OpenAI says GPT-6.1 Sol and GPT-6 Astra remain the better choice.
Can GPT-6.1 Terra fill out forms and enter invoices?
OpenAI lists filling out forms as a target task. It also gives the example of reading incoming invoices and entering their details into accounting software. Test it on your own documents, because scanned or messy invoices can behave differently from benchmark tasks.
What are the alternatives to GPT-6.1 Terra?
The main model alternatives are GPT-6.1 Sol, Claude Opus 5.5, Claude Sonnet 5.5, Kimi K3.1 Pro, and DeepSeek V4.1 Flash. For turning office files into finished reports and slides, Powerdrill Bloom offers a Free plan and Pro at $13.27/mo.
Sources: OpenAI, Introducing GPT-6.1 Terra · ChatGPT release notes · GPT-6.1 Terra system card addendum · OpenAI API pricing · OpenRouter, GPT-6.1 Terra · Claude API pricing · Kimi API pricing · DeepSeek API pricing. Details read from vendor pages on October 27, 2026.