GPT-6.1 Terra arrived at $1 per million input tokens, half the price of GPT-5.6 Terra
OpenAI says the new model matches GPT-6 Astra at about a tenth of the price and finishes tasks more than three times faster.
In brief
OpenAI's Terra now costs half as much
GPT-6.1 Terra costs $1 per million input tokens, 50% below its predecessor. OpenAI says it is over three times faster than GPT-6 Astra. Those figures are OpenAI's own.
OpenAI released GPT-6.1 Terra on October 26, 2026. It replaces GPT-5.6 Terra as the company's mid-size, fast and low-cost model. The details come from OpenAI's announcement, checked against coverage from Seeking Alpha and ithinkdiff.
Where you can use it
It is rolling out to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. In the API, developers can use it under the name gpt-6.1-terra. One outlet, Blockchain.News, also lists a Business plan. OpenAI's post and Seeking Alpha name only the four plans above.
The price
Standard API pricing is $1 per million input tokens, $5 per million output tokens and $0.05 per million cached input. A token is a small chunk of text, roughly a word fragment. Cached input means text the model has already seen and is reused.
OpenAI says this is 50% below GPT-5.6 Terra. The OpenAI-based report adds that it is also half the price of GPT-6.1 Sol, which costs $2 input and $10 output.
Speed and quality
OpenAI says Terra matches GPT-6 Astra at about a tenth of the price per token. It also says Terra completes tasks more than three times faster than Astra and GPT-6.1 Sol. These figures come from OpenAI's own timing runs.
On OpenAI's internal test of prompts designed to trigger errors, the error rate at low effort fell from 53.6% to 16.7%. At max effort it was 4.3%.
On Terminal-Bench 4.0, where an AI fixes problems from the command line, Terra solved 50.3% of tasks at max effort. GPT-5.6 Terra scored 21.5%.
Why it matters
OpenAI pitches Terra for visual reasoning, computer use and clearer writing. Imagine an AI that reads incoming invoices and types the details into accounting software. Cheaper and faster models make that kind of repetitive job cheaper to run at volume.
The context window, meaning how much text it can hold in mind at once, is 1,050,000 tokens. The output limit is 128,000 tokens.
Limits
Nearly every performance figure here was reported by OpenAI, not independently tested. The OpenAI-based report also says GPT-6.1 Sol and GPT-6 Astra are still the stronger options for the most complex coding work. Early reports that called Terra a leaked, unreleased model predate the launch.