OpenAI launches GPT-6.1 Terra: faster computer use at half the price, aimed at everyday work

OpenAI on October 26 released GPT-6.1 Terra, positioned as the everyday workhorse of its model lineup and the successor to GPT-5.6 Terra. In OpenAI's words, it is a fast and efficient model combining detailed visual reasoning and precise computer use with clearer, more natural writing, built for the communication and computer-based tasks that fill the workday: reading what is on screen, acting across apps, drafting proposals, interpreting technical drawings, filling out forms and editing videos.

Price cut in half: $1 per million input tokens

Pricing is the headline of this launch. Standard API rates are $1 per million input tokens, $5 per million output tokens and $0.05 per million cached input tokens, 50% below GPT-5.6 Terra and half the price of the flagship GPT-6.1 Sol. OpenAI also says usage allowances in ChatGPT Work and Codex go further with Terra than with Sol or Astra; developers can call it via the API as gpt-6.1-terra, and Amazon announced Bedrock availability the same day.

Strong vendor-reported benchmarks, not yet independently verified

On performance, OpenAI's own tests show 50.3% on the Terminal-Bench 4.0 coding benchmark at maximum reasoning effort, well above GPT-5.6 Terra's 21.5%; 15 percentage points higher than its predecessor on Agents' Last Exam, a professional software-use test, at about a third of the cost; and an error rate on its internal difficult-prompts test falling from 53.6% to 16.7% at low reasoning effort and to 4.3% at maximum effort. OpenAI also says Terra finishes tasks more than three times faster than Astra and Sol, with a context window of 1.05 million tokens. All of these figures are vendor-reported and have not been independently verified.

In positioning, OpenAI still reserves the hardest coding and deep research for GPT-6.1 Sol and GPT-6 Astra; Terra is the cheaper, faster option for high-volume everyday tasks. As labs push the price war into everyday models, the per-task cost of using AI for routine work is falling fast.