OpenAI launched GPT-6.1 Terra on October 26, a faster medium-tier model built for computer use, at standard API prices 50% lower than the GPT-5.6 Terra it replaces. According to OpenAI, the new model combines detailed visual reasoning and precise computer use with clearer, more natural writing, and it is available now in ChatGPT Work, Codex and through the API. OpenAI says Terra finishes computer tasks more than three times faster on average than GPT-6 Astra and GPT-6.1 Sol, and matches Astra at one-tenth the price per token. The launch completes the GPT-6 family refresh that left Terra out last month, when Sol and Luna arrived first.
What GPT-6.1 Terra Does
OpenAI pitches Terra as the model for the workday tasks that mix seeing a screen with acting on it. The company lists drafting proposals, interpreting technical drawings, filling out forms and editing videos among the jobs it handles. In ChatGPT Work and Codex, OpenAI says, a user’s usage allowance goes further with GPT-6.1 Terra than with GPT-5.6 Terra, and far more than with GPT-6.1 Sol or GPT-6 Astra.
Terra sits second from the bottom in OpenAI’s four-class lineup, above Luna and below Sol and Astra. OpenAI’s own description calls it a fast, efficient model for visual reasoning, precise computer use and clear, natural writing. For in-depth research or complex coding, the company still points users to GPT-6.1 Sol or GPT-6 Astra instead.
Price And Availability
Standard API pricing for GPT-6.1 Terra is $1 per million input tokens and $5 per million output tokens, according to OpenAI’s model documentation, with cached input at $0.05 per million tokens. That is half the standard price of GPT-5.6 Terra. By comparison, GPT-6.1 Sol costs $2 per million input tokens and $10 per million output tokens, while flagship GPT-6 Astra costs $10 and $50.
Terra is rolling out now to all Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. Developers can call it through the OpenAI API as gpt-6.1-terra. Its documented context window is 1.05 million tokens, with a maximum output of 128,000 tokens and a knowledge cutoff of June 1, 2026.
How It Measures Up
On OpenAI’s internal benchmark of error-inducing prompts, the error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra, a drop of nearly 37 percentage points. At maximum effort, OpenAI says, the error rate fell by 18 percentage points to 4.3%, nearly matching the rates for GPT-6 Astra and GPT-6.1 Sol.
OpenAI also says GPT-6.1 Terra substantially outperforms GPT-5.6 Terra on Agents’ Last Exam, its test of how well agents use software at work. The company frames the model as matching Astra at one-tenth the price per token while finishing tasks over three times faster on average than both Astra and GPT-6.1 Sol. Those figures are OpenAI’s own measurements, not independent tests.
Why It Matters
Terra matters because computer use is where everyday AI work actually happens: forms, documents, drawings and desktop software, rather than abstract reasoning tests. A model that handles those jobs at half the previous price, with fewer errors on OpenAI’s own tests, lowers the cost of automating routine office and development workflows. That is the same audience already adopting ChatGPT Work across large workforces, and the same developers building agents in Codex.
The launch also fills the gap in OpenAI’s lineup. When GPT-6 models for Sol and Luna arrived last month, Terra stayed on GPT-5.6, leaving the middle tier a generation behind. GPT-6.1 Terra closes that gap and gives developers a cheaper default for high-volume agent loops, while Sol and Astra keep the hardest reasoning and coding work. The trade-off to watch is whether the speed and price gains hold up outside OpenAI’s benchmarks, where independent testing has yet to weigh in.
Background
OpenAI’s model classes run from Luna at the cheapest end through Terra and Sol to Astra at the top, each balancing price against capability. GPT-6.1 Sol arrived earlier as the near-flagship option for complex work, and GPT-6.1 Sol has since been put to specialised jobs such as legacy-code migration. Terra’s previous version, GPT-5.6 Terra, had already received a price cut in July, and its retirement from GitHub Copilot’s model list in November underlined how quickly the lineup turns over.
Source: OpenAI: “GPT-6.1 Terra launch“
0 Comments