OpenAI (opens in new tab) released GPT-6.1 Terra on Monday, an upgrade to the middle model in its lineup that it says costs developers 50% less than GPT-5.6 Terra and handles computer tasks faster and more accurately, according to Decrypt (opens in new tab). OpenAI (opens in new tab)’s announcement calls it a significant step up from GPT-5.6 Terra and says its standard API prices are 50% lower. The Verge (opens in new tab) reported that the model is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and Decrypt (opens in new tab) said developers can reach it as gpt-6.1-terra. OpenAI (opens in new tab) describes it as “a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing,” according to The Verge (opens in new tab).
Decrypt (opens in new tab) reported that free users get the lower-tier GPT-5.6 Luna and that, for in-depth research or complex coding, OpenAI (opens in new tab) still points users to GPT-6.1 Sol or GPT-6 Astra.
On an internal OpenAI (opens in new tab) test built from prompts designed to trick models into making mistakes, the rate of factual errors at low reasoning effort fell from 53.6% for GPT-5.6 Terra to 16.7%, Decrypt (opens in new tab) reported. At maximum effort the rate fell 18 points to 4.3%. On Terminal-Bench 4.0, a test of coding agents working through a command-line terminal, Terra at maximum effort solved 50.3% of tasks against 21.5% for GPT-5.6 Terra, according to Decrypt (opens in new tab). OpenAI (opens in new tab) also says Terra finishes tasks more than three times faster on average than GPT-6 Astra and GPT-6.1 Sol. “All of the figures are OpenAI (opens in new tab)’s own, and the error test is an internal one,” Decrypt (opens in new tab) reported.
Plotly (opens in new tab), a data-visualization software company, ran the model through its private data-analytics exam on OpenRouter at the default reasoning setting and reported that it passed 40 of 43 questions (93%) for $1.31 for the whole exam, half the cost of GPT-6.1 Sol, which scored 43. A typical question took 15 seconds, Plotly (opens in new tab) said, making Terra the fastest of the 35 models on its leaderboard to score 40 or more. Plotly (opens in new tab)’s exam uses fabricated data from a synthetic wind-energy fleet, and the company said it keeps the exam private.
Plotly (opens in new tab) listed OpenRouter pricing for the model at $1.00 per million input tokens and $5.00 per million output tokens, which it said is half or less of what GPT-5.6 Terra cost.