OpenAI has launched GPT-6.1 Terra, a model built for everyday computer work: filling in forms, reading technical drawings, drafting proposals and editing video. API pricing drops to $1 per million input tokens and $5 per million output tokens, half the price of GPT-5.6 Terra and GPT-6.1 Sol. According to figures released by the company, agent tasks run about three times faster.
The complication comes from OpenAI's own testing. According to BleepingComputer and basic-tutorials.com, the company acknowledges the model is also less reliable in some scenarios, while remaining competitive on performance. The lower cost therefore comes with a trade-off that anyone automating sensitive tasks needs to weigh carefully.
Image: OpenAI — https://deploymentsafety.openai.com/gpt-6-1-terra
On the technical side, Terra accepts text and images as input and returns only text. The context window reaches 1,050,000 tokens, with up to 922,000 input and 128,000 output. On BenchLM the model ranks sixth out of 134 for tool use and computer tasks, with a category score of 76.4 out of 100; these are third-party evaluations, to be read as a relative indication.
In OpenAI's lineup Terra sits below Sol for balanced work and above Luna for high-volume tasks. The price comparison with rivals remains unfavourable for it, though: OpenRouter describes it as suited to everyday agentic work and high-volume loads, not as the choice for the hardest tasks.
Image: OpenAI — https://deploymentsafety.openai.com/gpt-6-1-terra
Available from today in the API as gpt-6.1-terra and in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu subscribers, the model shifts the comparison from raw power to cost per task. Teams running agents on repetitive work will need to check on their own workflows whether the savings hold up in reliability too.
No comments yet. Be the first!