OpenAI's GPT-6.1 Terra model

OpenAI launched GPT-6.1 Terra on October 26, a fast mid-range model priced at $1 per million input tokens and $5 per million output tokens. That is roughly half the cost of GPT-5.6 Terra and GPT-6.1 Sol, and about one tenth the cost of GPT-6 Astra. The model targets computer use, visual reasoning, coding, and business automation rather than being OpenAI's most capable offering. OpenAI engineer Ted Sanders described Terra as better than Astra and Sol at technical communication, comparable on vision and computer use but much faster, and somewhat weaker on most other tasks. On an internal coding benchmark, Terra scored 50.3% compared to 21.5% for predecessor GPT-5.6 Terra, with hallucination error rates dropping from 53.6% to 16.7% at low reasoning effort and to 4.3% at maximum reasoning effort. Terra performs on par with GPT-6 Astra on certain professional agent benchmarks while costing one tenth as much per token and running more than 3x faster than both Astra and GPT-6.1 Sol. Simon Willison positioned Terra as analogous to Anthropic's Sonnet tier, noting that sitting between the cheapest and most expensive options makes use-case selection less obvious. Willison found that max-reasoning use can still be slow and costly at the task level, with a single pelican benchmark run costing 11.3 cents and taking over four minutes, and characterized the release as not a bombshell for subscribers, noting it may help those hitting usage limits or wanting faster computer use.