GPT‑6.1 Terra targets high-volume visual reasoning and computer use at lower cost
OpenAI released GPT‑6.1 Terra on October 26 as a cost-efficient model for visual reasoning, precise computer use and high-volume professional work, with API support for image input, structured output, computer use, MCP and tool search.
Why it matters
A generative UI runtime needs a model to understand screens, choose actions and return verifiable structured state. Terra puts those capabilities into a lower-price, long-context tier for batch forms, screen operation and dynamic-interface backends, but a lower unit price does not establish reliable end-to-end workflows.
A closer look
Official documentation lists a 1,050,000-token context window, 128,000 maximum output tokens, and standard prices of $1 per million input tokens, $0.05 cached input and $5 output. Above 272K input tokens, the full request uses 2x input and cache prices and 1.5x output pricing. Through the Responses API it supports function calling, structured outputs, computer use, MCP, tool search, hosted shell, code interpreter and skills. OpenAI reports relative gains over the prior Terra model on Agents’ Last Exam, MMMU-Pro and OSWorld 2.1, plus 50.3% on Terminal-Bench 4.0 at maximum reasoning, but publishes no directly comparable absolute OSWorld score in the launch article.
Who this is for
Product and platform teams choosing models and runtime boundaries for generative UI, computer-use agents, batch forms and visual workflows
Original sources
- OpenAI official launch, model documentation and system card ↗openai.com
- GPT-6.1 Terra system card addendum ↗deploymentsafety.openai.com
- Official API model reference ↗developers.openai.com
Revisions & corrections (1)
— Official launch, API reference and system-card addendum establish the date, capabilities, pricing and safety boundaries. Claims are attributed to OpenAI, and unavailable absolute OSWorld or generative-UI-specific results are explicitly excluded.