OpenAI Launches GPT-6.1 Terra Model with 50% Price Cut
OpenAI has released GPT-6.1 Terra, a new medium-sized AI model designed for speed and cost-efficiency. Priced 50% lower than its predecessor GPT-5.6 Terra, the model features improved visual reasoning and computer-use capabilities. It aims to balance intelligence with affordability for high-volume automation tasks, marking a significant shift in OpenAI's tiered model strategy.
Key points
- OpenAI, the artificial intelligence company, has introduced GPT-6.1 Terra, a new medium-sized model class positioned between its high-cost Astra and Sol models.
- The GPT-6.1 Terra model is priced at 50% lower than the previous GPT-5.6 Terra version, offering a significant cost reduction for developers and enterprise users.
- Performance benchmarks indicate a drop in error rates from 53.6% to 16.7% on internal tests, while scoring 15 percentage points higher than GPT-5.6 on the Agents' Last Exam benchmark.
- The model excels in visual reasoning and precise computer-use tasks, such as interpreting technical drawings, drafting proposals, and automating data entry for invoices.
- Users of ChatGPT Work and Codex can now perform more tasks within the same usage allowance, with Terra providing greater utility than the more expensive GPT-6.1 Sol or Astra models for standard workflows.
OpenAI Expands Model Lineup with Cost-Efficient Terra Update
OpenAI has officially released GPT-6.1 Terra, a new iteration of its medium-sized artificial intelligence model. This update replaces the previously available GPT-5.6 Terra model, marking a strategic shift in the company's approach to balancing computational power with affordability. The release is part of OpenAI's broader portfolio, which now includes four distinct model classes: Astra, Sol, Terra, and Luna, ranging from the highest capability and cost to the lowest. The introduction of GPT-6.1 Terra aims to provide a more efficient solution for users who require robust performance without the premium pricing associated with the company's top-tier models.
Significant Price Reduction and Performance Gains
A central feature of the GPT-6.1 Terra announcement is its pricing structure. OpenAI states that the standard API prices for the new model are 50% lower than those of the GPT-5.6 Terra it replaces. This cost reduction is designed to make high-volume automation tasks more practical for developers and enterprises. For instance, tasks such as reading incoming invoices and entering details into accounting software can now be performed at a fraction of the previous cost. Additionally, the pricing is significantly lower than that of the GPT-6.1 Sol model, positioning Terra as a cost-effective alternative for routine workloads.
In terms of performance, OpenAI reports substantial improvements over the previous generation. Internal benchmarks measuring error-inducing prompts show that the error rate at low reasoning effort has dropped from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra. This represents a decrease of nearly 37 percentage points. Furthermore, on the Agents' Last Exam, a benchmark of professional tasks that includes turning architectural drawings into 3D models, GPT-6.1 Terra scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost. At maximum reasoning effort, the model's performance is noted to be over three times more efficient in specific metrics compared to its predecessor.
Enhanced Capabilities for Computer Use and Visual Reasoning
The GPT-6.1 Terra model is specifically optimized for communication and computer-based tasks that typically fill a workday. OpenAI highlights its ability to combine detailed visual reasoning with precise computer use. This makes the model suitable for a variety of applications, including drafting proposals, interpreting technical drawings, filling out forms, and editing videos. The model's proficiency in visual understanding allows it to review complex media, such as creating time-coded logs for film restoration teams by analyzing animated shots.
OpenAI also notes that GPT-6.1 Terra performs on par with the GPT-6 Astra model at only one-tenth the price per token for certain tasks. This comparison underscores the model's efficiency, suggesting that users do not need to invest in the most expensive model for many standard professional tasks. The improved natural writing capabilities also contribute to clearer communication, making the model a versatile tool for both technical and creative workflows.
Impact on ChatGPT Work and Codex Users
For users of ChatGPT Work and Codex, the introduction of GPT-6.1 Terra means that their existing usage allowance will cover more tasks than before. The same allowance that previously supported fewer operations with GPT-5.6 Terra now enables a higher volume of work. This benefit extends to users who previously relied on GPT-6.1 Sol or GPT-6 Astra, as Terra offers a more economical path for high-volume tasks without a significant sacrifice in capability. The update is available immediately, allowing users to integrate the new model into their existing workflows to enhance productivity and reduce costs.
Key takeaways
- OpenAI has released GPT-6.1 Terra, a new medium-sized AI model replacing GPT-5.6 Terra.
- The new model is priced 50% lower than the previous version, reducing costs for high-volume automation.
- Error rates on internal benchmarks have dropped significantly, from 53.6% to 16.7% at low reasoning effort.
- GPT-6.1 Terra excels in visual reasoning and computer-use tasks, such as interpreting technical drawings.
- Users of ChatGPT Work and Codex can perform more tasks within the same usage allowance.
- The model offers performance comparable to higher-tier models at a fraction of the cost.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.
