OpenAI launches GPT-6.1 Terra for faster and cheaper daily tasks

A balance of speed and cost

OpenAI recently introduced GPT-6.1 Terra, a new artificial intelligence model designed to streamline everyday work tasks. The company developed this model to provide a better balance between speed, performance, and cost for its users. It aims to handle practical needs like drafting proposals, reading technical drawings, and filling out forms.

According to the announcement by Digit on MSN, the model is positioned as a significant step up from the previous GPT-5.6 Terra. It is specifically built to be faster and more affordable for high-frequency tasks. This makes it a practical choice for businesses that need to process large amounts of data or content without high overhead.

The model is now accessible to a wide range of users, including those with Plus, Pro, Enterprise, and Edu subscriptions through ChatGPT Work and Codex. Developers can also integrate the model into their own applications using the OpenAI API under the specific name gpt-6.1-terra.

Visual and computer capabilities

One of the primary features of GPT-6.1 Terra is its enhanced ability to understand visual information and interact with computer applications. This allows the model to perform complex actions that require seeing a screen or a document. For example, it can create 3D models based on architectural drawings provided by users.

OpenAI also tested the model on specialized media tasks, such as reviewing animated films to create time-coded reports. These capabilities demonstrate the model's move toward becoming a more versatile assistant for creative and technical industries. It can bridge the gap between static text and dynamic visual media.

The model's performance in these areas is backed by specific testing results. OpenAI reported that GPT-6.1 Terra scored 15 percentage points higher than the GPT-5.6 Terra on the Agents' Last Exam benchmark when using medium reasoning effort. This improvement was achieved while only costing one-third of the previous model's price.

Accuracy and coding improvements

OpenAI has worked to reduce factual errors within the GPT-6.1 Terra model to make it more reliable for professional use. On internal benchmarks, the error rate at low reasoning effort dropped from 53.6 percent with the older GPT-5.6 Terra to just 16.7 percent. At the highest level of reasoning effort, the error rate fell even further to 4.3 percent.

The model also shows notable progress in software development and coding tasks. On the Terminal-Bench 4.0 coding benchmark, it successfully solved 50.3 percent of tasks at maximum reasoning effort. This represents more than double the 21.5 percent score that was published for the GPT-5.6 Terra model.

Despite these gains, OpenAI clarified that the model is not intended to replace every existing tool in their lineup. The company noted that GPT-6.1 Sol and GPT-6 Astra remain the better choices for the most complex coding tasks. This suggests that Terra is optimized for a specific middle ground of speed and utility.

Pricing and API availability

For developers and large-scale users, the pricing structure of GPT-6.1 Terra is a key component of its release. The company priced the API at $1 per million input tokens and $5 per million output tokens. They also offer a rate of $0.05 per million cached input tokens for more efficient data handling.

These prices represent a significant reduction in cost for those moving away from older models. OpenAI stated that the standard API pricing for this new model is 50 percent lower than that of both GPT-5.6 Terra and GPT-6.1 Sol. This price drop is intended to make the model more accessible for high-volume automated workflows.

By lowering the barrier to entry, OpenAI is making it easier for companies to integrate advanced AI into their daily operations. Users can now deploy a model that handles visual understanding and computer use at a fraction of the previous cost. This shift allows for more experimentation with AI agents in real-world business environments.