GPT-6.1 Terra lands on API and Codex

In this article you will see:

• What GPT-6.1 Terra does with images and interfaces
• Where the model is available and what its identifier is in the API
• What the tests published by OpenAI show
• How to access the model for visual automations

GPT-6.1 Terra combines visual reasoning and computer use

OpenAI positions Terra for tasks that connect perception and action, such as interpreting documents and technical drawings or operating application screens. The proposal goes beyond answering questions about an image.

The card identifies the official model page. The documentation lists support for computer use and other tools via Responses API, which allows the model to act in flows defined by the application.

GPT-6.1 Terra arrives at API, Codex and ChatGPT Work

From the announcement, GPT-6.1 Terra is available in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu accounts. Developers can call it via API with identifier gpt-6.1-terra.

The API sheet lists a context window of 1,050,000 tokens and maximum output of 128,000 tokens. For computer tool use, OpenAI advises using Responses API.

OpenAI also disclosed a standard price of US$ 1 per million input tokens and US$ 5 per million output. Cached input costs US$ 0.05 per million tokens. These values are part of the announcement and may change, so check the current API pricing page.

Those who follow the arrival of OpenAI models to other environments can also read the announcement about Codex models and flows on Google Cloud. For API context, see how OpenAI expanded prompt cache.

Tests published by OpenAI point to gains in visual tasks

In the Agents’ Last Exam benchmark, OpenAI reports that GPT-6.1 Terra scored 15 percentage points above GPT-5.6 Terra with medium reasoning effort, for about one third of the cost per task.

At maximum effort, the company says Terra completes tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol. These are results disclosed by OpenAI itself, not an independent evaluation done by bridgecode.

How to access GPT-6.1 Terra today

In the API, select gpt-6.1-terra. To combine the model with computer tools, use the Responses API and official model documentation.

Before automating tasks in interfaces, define which actions the agent can execute and when it needs approval. The model makes it possible to interpret screens and act, but the application remains responsible for controlling scope and permissions.