GPT-6.1 Terra arrived with a blunt proposition: make AI agents useful enough for the workday without making every automated task prohibitively expensive. A report on the launch described the model as available immediately to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and repeated OpenAI’s claim that it combines visual reasoning, computer use and more natural writing.¹

OpenAI’s fuller announcement places the model between raw capability and operating cost. The company says Terra’s API pricing is 50% below both GPT-5.6 Terra and GPT-6.1 Sol, aimed at repetitive, high-volume jobs such as extracting invoice details and entering them into accounting software.² On its Agents’ Last Exam benchmark, OpenAI says Terra scored 15 percentage points above its predecessor at roughly one-third of the cost, while approaching GPT-6 Astra’s performance at one-tenth the token price.²

The pitch is not merely cheaper automation, but speed. OpenAI says Terra completed tested computer-use tasks more than three times faster than Astra and Sol, and reported a sharp reduction in errors on its internal adversarial prompts.² It also says the model more than doubled GPT-5.6 Terra’s published Terminal-Bench 4.0 result at maximum reasoning effort—though it concedes Sol and Astra remain preferable for the most complex coding work.²

That caveat has not dampened executive enthusiasm. Greg Brockman wrote that Terra’s computer use had crossed “a qualitative threshold” at which it “feels faster at using a computer than what i can do unassisted.”Greg Brockman (gdb)³ Sam Altman called it “a new era of computer use.”Sam Altman (sama)⁴

For now, those are company claims anchored in company benchmarks. But the economics are the point: OpenAI is betting that faster, cheaper agents will matter as much as frontier-model supremacy in turning AI from a demonstration into routine office infrastructure.