OpenAI Launches GPT-6.1 Terra With API Pricing 50% Lower Than GPT-5.6 Terra
OpenAI has released GPT-6.1 Terra, a faster and cheaper AI model designed for using computers, understanding images and other visual information, writing code, and handling everyday office tasks. The company announced the release in a blog post on October 26.
The model is now available to Plus, Pro, Enterprise, and Edu subscribers through ChatGPT Work and Codex. Developers can also use it through the API with the model name gpt-6.1-terra.
OpenAI says Terra is replacing GPT-5.6 Terra. Its standard API prices are 50% lower than those of both GPT-5.6 Terra and GPT-6.1 Sol. The price is $1 for every million input tokens, $0.05 for every million cached input tokens, and $5 for every million output tokens. Tokens are small pieces of text or other data that the model processes.
OpenAI says Terra is designed for daily work such as writing proposals, understanding technical drawings, filling in forms, and editing videos. For example, it can read incoming invoices and enter the information into accounting software.
GPT-6.1 Terra Benchmark Results and Error Rates
On Agents' Last Exam, a test of professional tasks that involve using a computer, GPT-6.1 Terra scored 15 percentage points higher than GPT-5.6 Terra when both used a medium reasoning setting. These tasks include turning architectural drawings into 3D models. According to OpenAI, Terra achieved this result while costing about one-third as much. The company also says Terra performs about as well as GPT-6 Astra while costing one-tenth as much per token. It completes tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol, according to OpenAI.
OpenAI also compared Terra with Anthropic's Claude Opus 5.5. At the highest reasoning setting, the company says Terra uses more than three times fewer tokens to do the same work and costs one-fourth as much per token. Together, these differences mean that a task costs about nine times less with Terra, according to OpenAI.
Terra also made fewer factual mistakes in OpenAI's internal test, which used prompts designed to cause errors. At a low reasoning setting, the error rate dropped from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra. At the highest setting, the error rate fell by 18 percentage points to 4.3%. This is close to the error rates of GPT-6 Astra and GPT-6.1 Sol.
Terra also did better at coding. On Terminal-Bench 4.0, a test of AI agents that write code using command-line tools, Terra completed 50.3% of tasks at the highest reasoning setting. GPT-5.6 Terra completed 21.5%. OpenAI still recommends GPT-6.1 Sol or GPT-6 Astra for the hardest coding tasks.
What ChatGPT and API Users Should Check
For people who use ChatGPT Work and Codex, the main benefit is that they can complete more tasks within the same usage allowance. OpenAI says GPT-6.1 Terra allows users to do more work than GPT-5.6 Terra, and much more than GPT-6.1 Sol or GPT-6 Astra.
Developers who run large numbers of automated tasks can change their model identifier to gpt-6.1-terra. Before switching their live systems, they can compare how much each completed task costs with Terra and their current model.
For safety, OpenAI says GPT-6.1 Terra meets the Critical level for cybersecurity risk under its Preparedness Framework. It also keeps the safety protections developed for GPT-6 Astra.
OpenAI also says Terra is better than GPT-5.6 Terra at resisting prompt-injection attacks, which try to trick an AI into ignoring its instructions. The company published more details in the model's system card. The launch comes after OpenAI recently expanded ChatGPT ads to 89 markets.
All benchmark results come from tests run by OpenAI itself. The company has not said whether GPT-6.1 Terra will become available on ChatGPT's free plan.