OpenAI’s GPT-6.1 Terra: A Faster, More Affordable Leap in Computer Use

• Cost Efficiency: GPT-6.1 Terra is priced 50% lower than GPT-5.6 Terra and GPT-6.1 Sol, with standard API rates at $1 per million input tokens and $5 per million output tokens.
• Speed Advantage: The model completes computer-use tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol.
• Performance Gains: On the Agents’ Last Exam benchmark, it scored 15 percentage points higher than GPT-5.6 Terra at medium reasoning effort for roughly one-third of the cost.
• Coding Improvements: Terminal-Bench 4.0 scores rose to 50.3% at maximum reasoning effort, up from 21.5% for GPT-5.6 Terra.
• Reduced Errors: Factual error rates dropped significantly from 53.6% to 16.7% at low reasoning effort.
• Accessibility: Available to Plus, Pro, Enterprise, and Edu users via ChatGPT Work and Codex, and to developers via the API identifier gpt-6.1-terra.

GPT-6.1 Terra delivers substantial improvements in visual understanding, factual accuracy, and software engineering capabilities. Its performance is particularly notable on benchmarks that test professional tasks involving both code and visual data.

On the Agents’ Last Exam, which evaluates proficiency in software and visual information processing, GPT-6.1 Terra outperformed its predecessor, GPT-5.6 Terra, by 15 percentage points at medium reasoning effort. This improvement was achieved at approximately one-third of the cost. Furthermore, it matched the performance of GPT-6 Astra on this specific benchmark while costing only about one-tenth per token.

Coding capabilities have also seen a dramatic upgrade. In tests using Terminal-Bench 4.0, which assesses coding agents on command-line tasks, GPT-6.1 Terra achieved a score of 50.3% at maximum reasoning effort. This is a significant increase from the 21.5% score recorded by GPT-5.6 Terra.

Despite these advances, OpenAI advises that for the most demanding coding and research tasks, users should still consider GPT-6.1 Sol or GPT-6 Astra. However, for general professional automation, the error reduction is striking. Internal tests show that factual errors at low reasoning effort fell from 53.6% to 16.7%, making it a more reliable tool for routine digital work.

The primary appeal of GPT-6.1 Terra lies in its affordability and efficiency. OpenAI has set the standard API prices at US$1 per million input tokens, US$0.05 per million cached input tokens, and US$5 per million output tokens. These rates represent a 50% reduction compared to GPT-5.6 Terra and GPT-6.1 Sol.

The model’s token efficiency is also superior to competitors. OpenAI reports that GPT-6.1 Terra is over three times as token-efficient as Anthropic’s Claude Opus 5.5 model, while costing only one-fourth of its price per token. This combination results in approximately nine times lower cost per task compared to previous iterations or rival models.

For businesses looking to scale AI-powered workflows, these pricing structures make automating repetitive digital tasks and processing large volumes of documents more practical than ever. The model is accessible to Plus, Pro, Enterprise, and Edu users through ChatGPT Work and Codex. Developers can integrate it into custom applications using the OpenAI API with the specific model identifier gpt-6.1-terra.

This launch is supported by a robust network of strategic partners who have invested heavily in OpenAI’s infrastructure:

• Microsoft: As a core partner, Microsoft holds a 27% stake in the newly structured OpenAI Group Public Benefit Corporation (PBC) following a historic $13 billion initial investment.
• AWS: Amazon Web Services serves as a strategic multi-cloud partner, collaborating on a stateful runtime environment for Amazon Bedrock after a $50 billion investment in OpenAI.
• NVIDIA: NVIDIA has been OpenAI’s essential hardware supplier for a decade, providing the GPUs that have powered ChatGPT since its inception.
• Broadcom: To reduce reliance on standard GPUs, OpenAI has partnered with Broadcom to co-develop custom AI accelerators and Ethernet networking racks.

GPT-6.1 Terra represents a pivotal moment in the evolution of practical AI. By combining faster execution speeds, significantly reduced costs, and improved accuracy in visual and coding tasks, it lowers the barrier for professionals to automate their daily workflows. Whether through direct user interfaces or developer APIs, this model offers a powerful, cost-effective solution for those looking to enhance productivity without the premium price tag of previous generations.