OpenAI on Tuesday introduced GPT-6.1 Terra, a new model aimed at making workplace AI agents cheaper to run without sacrificing their ability to see, reason and operate software. The company said Terra combines “detailed visual reasoning and precise computer use” with more natural writing, targeting jobs from proposal drafting and technical-drawing interpretation to form filling and video editing.¹

The financial argument is central. OpenAI said standard API prices are 50% below GPT-5.6 Terra and GPT-6.1 Sol, with input tokens priced at $1 per million and output tokens at $5 per million. On its own Agents’ Last Exam benchmark, it said Terra scored 15 points above GPT-5.6 Terra at medium reasoning effort for roughly one-third of the cost; it also claimed the model completed tasks more than three times faster than GPT-6 Astra and Sol.¹

The rollout begins immediately for Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, while developers can call the model through the API. OpenAI also says it has reduced factual mistakes: on an internal error-inducing prompt test, its low-effort error rate fell from 53.6% in the prior Terra model to 16.7%.¹

Company leaders framed the release as more than an incremental speed boost. Greg Brockman wrote that computer use with Terra had crossed “a qualitative threshold” at which it feels faster than his own unassisted computer work.Greg Brockman (gdb)² Sam Altman likewise called it “this new era of computer use.”Sam Altman (sama)³

Outside the launch post, coverage echoed OpenAI’s description rather than independently validating the performance claims, reporting that the model is available now and that OpenAI says it excels at the communication and computer-based tasks of the workday.⁴ The company itself draws a line around the hype: for the most complex coding assignments, it says GPT-6.1 Sol and GPT-6 Astra remain the better choice.¹