OpenAI launches GPT-6.1 Terra: efficient and multifunctional

OpenAI has recently presented the new AI model GPT-6.1 Terra. This model, which falls under the mid-range category in OpenAI's offerings, stands out due to its ability to view a screen and undertake actions independently, such as clicking and typing in various applications. GPT-6.1 Terra promises to be faster and cheaper than its predecessor, GPT-6.1 Sol, and offers Dutch users the opportunity to try the model directly via ChatGPT Work or the API, although specific performance in Dutch has not yet been confirmed.

One of the most notable features of GPT-6.1 Terra is its lower error rate. This has been reduced from 53.6 percent in an internal test with GPT-5.6 Terra to 4.3 percent with a high stakes reasoning application. This not only makes the model more accurate but also more powerful for various tasks, from drafting proposals to automating administrative work.

The use cases for GPT-6.1 Terra include tasks such as filling out forms, video editing, and automatically processing data from invoices. The model also proves to be more than three times as fast in completing tasks as its predecessors. This makes it suitable for agents who need to perform many repetitive actions.

For users, GPT-6.1 Terra is available through various subscriptions, including Plus, Pro, Enterprise, and Edu. The costs for using the API are $1 per million input tokens and $5 per million output tokens, which makes the price half as low as that of the earlier models. However, there are currently questions regarding support for the Dutch language and availability in the EU, which has not yet been clarified by OpenAI.

Read the full article from AI Insider.