OpenAI Launches GPT-6.1 Terra With Faster Computer Use and Lower API Costs

A Faster Model for Work on a Screen

OpenAI has introduced GPT-6.1 Terra, a model designed for visual reasoning and computer-based tasks, with standard API prices half those of GPT-5.6 Terra and GPT-6.1 Sol. The company says the model is available in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, as well as through its API.

Terra is aimed at tasks that require an AI system to interpret what is on a screen and take actions inside software. OpenAI lists examples ranging from filling out forms and editing video to interpreting technical drawings and drafting proposals. The announcement also points to clearer writing as an area of improvement over GPT-5.6 Terra.

For businesses evaluating AI tools, the practical question is whether an agent can complete a useful workflow reliably enough to justify the time and expense of supervising it. A faster, less expensive model could make more repetitive work economical to automate, although the results will depend on the task and the controls around it.

---

The Cost of Using Terra

The standard API price is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. Tokens are the units used to measure the text and other content processed by a model; an application's final bill depends on how much it sends, how much it receives and whether eligible input is cached.

OpenAI says those standard rates are 50% lower than the prices for GPT-5.6 Terra and GPT-6.1 Sol. It also says the same usage allowance in ChatGPT Work and Codex covers more tasks with Terra than with its predecessor, and substantially more than with Sol or GPT-6 Astra. Actual task costs can vary because different models may use different numbers of tokens to reach an answer.

That distinction matters for high-volume applications. A lower price per token is useful, but the total cost also depends on how often an agent retries a task, how long its responses are and how much human review is needed. OpenAI gives processing invoices and entering their details into accounting software as an example of work developers might automate.

---

What OpenAI's Tests Show

On Agents' Last Exam, a benchmark covering professional tasks, OpenAI says GPT-6.1 Terra scored 15 percentage points higher than GPT-5.6 Terra at medium reasoning effort while costing about one-third as much per task. The company says Terra performed on par with GPT-6 Astra in that setting, while also reporting a substantially lower price per token.

OpenAI says Terra completed tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol in that comparison. It also reported improvements on visual reasoning and desktop computer-use tests. These are company-reported benchmark results; a benchmark score does not guarantee the same speed or reliability across every application.

For coding, OpenAI reports a 50.3% score at maximum reasoning effort on Terminal-Bench 4.0, compared with a published 21.5% for GPT-5.6 Terra. The company still identifies GPT-6.1 Sol and GPT-6 Astra as better choices for its most complex coding tasks.

---

Factuality and Safety Still Matter

OpenAI also reported fewer factual errors on an internal set of prompts designed to induce mistakes. At low reasoning effort, the reported error rate fell from 53.6% for GPT-5.6 Terra to 16.7% for GPT-6.1 Terra. At maximum effort, OpenAI reported a 4.3% error rate for the new model.

Those figures describe a specific internal test rather than a universal error rate. Users should still check consequential outputs, particularly when an agent is entering information into business software or acting on instructions gathered from outside content.

OpenAI says GPT-6.1 Terra meets the Critical threshold for cybersecurity under its Preparedness Framework and retains protections developed for GPT-6 Astra. The company also reports stronger resistance to prompt injection, in which malicious instructions attempt to redirect an agent from its assigned task. Developers can access the model through the API under the name gpt-6.1-terra.