GPT-6.1 Terra

OpenAI presents GPT-6.1 Terra as a fast, efficient midrange model emphasizing visual reasoning, computer use, and clearer writing. Its intended tasks include drafting proposals, interpreting technical drawings, filling out forms, and editing video. Like Anthropic’s Sonnet, Terra sits between the cheapest and most capable models, making the choice of when to use it less obvious. OpenAI’s Ted Sanders says it is stronger than Astra and Sol at technical communication, roughly comparable on vision and computer use but much faster, and weaker at most other tasks. For API developers, it offers a less expensive alternative to Sol that substantially outperforms Luna in vision, computer control, and communication.

Terra is positioned as a practical option for repetitive automated tasks, where speed and lower cost matter; for subscribers, it is a modest addition that may help users stretch usage limits or get faster computer interactions. Its API price is $1 per million input tokens and $5 per million output tokens - nearly half the price of GPT-5.6 Terra and GPT-6.1 Sol, but about ten times Luna’s. A comparison using pelicans riding bicycles shows results across the reasoning levels of OpenAI’s current lineup. At maximum reasoning, Terra produced a correctly formed pelican and bicycle, matching the family’s color scheme; the run cost 11.3161 cents and took four minutes and eleven seconds.