OpenAI announced GPT-6.1 Terra on 26 October 2026, positioning the model not as a premium reasoning machine but as a workhorse for screen-based tasks: invoices, online forms, technical drawings and video editing. The company says it combines visual reasoning, natural-language writing and the ability to act across applications, while cutting token prices by 50% from its prior Terra generation.¹

The sales pitch is speed. OpenAI President Greg Brockman said computer use with Terra had crossed “a qualitative threshold” and felt faster than working unaided.Greg Brockman (gdb)² Chief executive Sam Altman framed the release as the beginning of “this new era of computer use.”Sam Altman (sama)³

Terra is now available to Plus, Pro, Enterprise and Edu users through ChatGPT Work and Codex. OpenAI describes it as “a fast, efficient model” built around detailed visual reasoning, precise computer use and clearer writing—the sort of communication and computer tasks that dominate an ordinary workday.⁴ Its standard API price is $1 per million input tokens and $5 per million output tokens; OpenAI also says the model is more than three times faster on computer-use tasks than GPT-6 Astra and GPT-6.1 Sol.¹

But the cheaper model lands in a less celebratory market debate. Ramp’s corporate-spending-based token price index put prices at about 59 cents per million tokens last Friday, down roughly 35% since late June. That decline points to a price war between OpenAI and Anthropic, even as AI usage rises.⁵

For OpenAI and AI boosters, lower costs are the mechanism that unlocks adoption at scale. For investors, the distinction is harsher: token use can surge while token spending—the revenue pool expected to support the data-centre buildout—fails to keep pace. Ramp’s recent figures, the report notes, appear to show spending growth flattening.⁵