OpenAI has introduced GPT-6.1 Terra, positioning the model as a cheaper, faster route to AI agents that can see a screen, reason about what they find and carry out work across desktop applications.¹

The company says Terra follows GPT-5.6 Terra with stronger visual reasoning, more precise computer use and more natural writing—aimed at routine office tasks such as interpreting technical drawings, completing forms, editing video and moving invoice details into accounting software. Standard API prices are 50% below both GPT-5.6 Terra and GPT-6.1 Sol, OpenAI says.¹

The central pitch is not simply lower pricing, but a shift in what automation can economically handle. On OpenAI’s Agents’ Last Exam benchmark, the company says Terra scored 15 percentage points above its predecessor at medium reasoning effort while costing about a third as much. It also says the model completed tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol.¹

Sam Altman framed the launch as an inflection point: “we are especially excited for this new era of computer use.”Sam Altman (sama)² Greg Brockman pushed the claim further, saying Terra’s computer use had crossed “a qualitative threshold where it feels faster at using a computer than what i can do unassisted.”Greg Brockman (gdb)³

OpenAI also argues Terra is more reliable, citing an internal error-inducing benchmark on which its low-effort error rate fell from 53.6% to 16.7%. But the company draws a line around the release’s limits: Terra reached 50.3% on Terminal-Bench 4.0 at maximum reasoning effort, more than double GPT-5.6 Terra’s published result, yet GPT-6.1 Sol and GPT-6 Astra “remain the better choice” for the most complex coding work.¹

Terra is available immediately to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and through OpenAI’s API at $1 per million input tokens and $5 per million output tokens.¹