OpenAI launches GPT-6.1 Terra for computer use at half the earlier price
By Anika SharmaAI & Innovations Editor, Peepals Tech
Representative image: Pexels / Pixabay
OpenAI released GPT-6.1 Terra on 26 October 2026, a mid-tier AI model built to operate computer screens, priced 50% below its predecessor, the company says.
What Happened
OpenAI released GPT-6.1 Terra, a mid-priced artificial intelligence model designed for "computer use", meaning tasks where the AI reads a screen and clicks, types and moves through software the way a person would, on Monday, 26 October 2026. The company says the model does this work better than its predecessor, GPT-5.6 Terra, while its price for developers is half as much. It is available to paying ChatGPT users on the Plus, Pro, Enterprise and Edu plans and to developers through OpenAI's API, the programming interface that lets other software use the model.
Key Facts
- Model: GPT-6.1 Terra, OpenAI's mid-tier model, positioned below its more expensive GPT-6.1 Sol.
- Launch: 26 October 2026, according to most reports.
- Developer price: 1 US dollar per million input tokens and 5 dollars per million output tokens, 50% below GPT-5.6 Terra and GPT-6.1 Sol, according to OpenAI. A token is a small piece of text, roughly three-quarters of an English word.
- Cheaper modes: batch and flex processing cost half the standard rate; a fast mode costs double.
- Who can use it: ChatGPT Plus, Pro, Enterprise and Edu subscribers, including in Codex, OpenAI's coding tool, and developers through the API.
- OpenAI's claim: on the Agents' Last Exam test of computer-use tasks, Terra at medium effort scores 15 percentage points higher than GPT-5.6 Terra at roughly a third of the cost.
- OpenAI's claim: on Terminal-Bench 4.0, a test of command-line tasks, it completes 50.3% of tasks at maximum effort, against 21.5% for GPT-5.6 Terra.
Why It Matters
The race among AI companies is moving from chatbots that answer questions to "agents" that carry out tasks: filling forms, moving data between programs, testing software or booking things online. Computer use is a key skill for such agents, because much of the world's work still happens in software built for humans, not for machines. A model that can reliably operate those screens could automate routine office work.
Cost is the main barrier. Agents take many steps for one task, and each step uses tokens, so expenses add up quickly. By cutting the price in half while claiming better results, OpenAI is betting that cheaper agents will be used far more widely. The company's message is about economics as much as raw intelligence: a mid-tier model that does most jobs well enough at a fraction of the price of its top model.
The performance figures come from OpenAI's own tests. Independent testers will need time to check whether the gains hold across ordinary tasks or mainly on the selected benchmarks, and AI agents still make mistakes when screens change or instructions are unclear. Businesses that hand real work to such agents usually keep a human checking the results.
For India, where many IT services and back-office firms do exactly the kind of screen-based work these agents target, cheaper computer-use models are both an opportunity and a pressure. Indian developers and start-ups building on OpenAI's API can now run agents at lower cost, while companies whose revenue depends on manual processing will face clients asking how much can be automated.
| Item | GPT-5.6 Terra | GPT-6.1 Terra |
|---|---|---|
| Input price, per million tokens | About 2 dollars | 1 dollar |
| Output price, per million tokens | About 10 dollars | 5 dollars |
| Terminal-Bench 4.0 (OpenAI figures) | 21.5% | 50.3% |
Impact
Short-term: Paying ChatGPT users can try the model now, and developers can switch their applications to it through the API at the lower price.
Long-term: If independent tests confirm OpenAI's claims, cheaper computer-use agents could speed up the automation of routine digital work, from data entry to software testing, and push rival AI companies to cut their own prices.
Who is affected: Developers and start-ups building AI agents, businesses that automate office tasks, IT services and back-office firms, ChatGPT subscribers, and competing AI model makers.
Key Takeaway
OpenAI's GPT-6.1 Terra, launched on 26 October 2026, is a mid-tier model for operating computer screens that the company says works better than its predecessor at half the developer price, though the gains are OpenAI's own claims.
Questions and Answers
What is GPT-6.1 Terra?
GPT-6.1 Terra is a mid-tier OpenAI model released on 26 October 2026, built mainly for computer use: reading a screen and operating software by clicking and typing, the way a person would.
How much does GPT-6.1 Terra cost?
For developers using OpenAI's API, it costs 1 US dollar per million input tokens and 5 dollars per million output tokens, which OpenAI says is 50% less than GPT-5.6 Terra. Batch and flex modes cost half that.
Can I use GPT-6.1 Terra in ChatGPT in India?
It is available to ChatGPT Plus, Pro, Enterprise and Edu subscribers, according to OpenAI. Free users were not listed among those with access.
Is GPT-6.1 Terra really better than earlier models?
OpenAI says it scores 15 percentage points higher than GPT-5.6 Terra on one computer-use test and completes 50.3% of tasks on Terminal-Bench 4.0 against 21.5% before. These are the company's own figures; independent tests have yet to confirm them.
Disclaimer: Prepared by the Peepals newsroom from publicly available sources with AI assistance. Information is accurate to the best of our knowledge at the time of publication and may change. Images may be representative. Not professional advice. Report an error via our contact page.
By Anika Sharma, AI & Innovations EditorCovers AI & Innovation and more for Peepals Tech. All stories by Anika Sharma
SonicWall opens Pune office, plans 400 engineering roles in India
AssetMark opens Hyderabad tech centre with room for about 500 staff
Get the day's stories in your inbox
Peepals Weekly: the week decoded, free. One email a week, unsubscribe in one click.
Anika Sharma is a pen name of the Peepals Global Editorial Team, which researches, fact-checks and publishes every story. The portrait is an illustration.
