OpenAI launches GPT-6.1 Terra with faster visual reasoning

OpenAI introduced GPT-6.1 Terra, a model designed to combine visual understanding, automated computer use and clearer natural-language output. According to OpenAI, the model targets everyday work tasks such as drafting documents, interpreting technical drawings, filling forms and editing video.

Capabilities and benchmarks

OpenAI reported that GPT-6.1 Terra outperforms GPT-5.6 Terra across multiple benchmarks. On Agents’ Last Exam, a professional-task benchmark, GPT-6.1 Terra scored 15 percentage points higher than GPT-5.6 Terra at roughly one-third the cost, and matched GPT-6 Astra’s performance at about one-tenth the price per token, the company said.

On MMMU-Pro (vision) and OSWorld 2.1 tests of desktop-app workflows, OpenAI said GPT-6.1 Terra beat GPT-5.6 Terra at every reasoning setting and surpassed GPT-5.6 Terra’s best OSWorld 2.1 score at a fraction of the task cost. OpenAI also reported that Terra completes tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol in some comparisons.

Factuality, coding, safety and availability

OpenAI stated that efforts to reduce hallucinations produced lower error rates for GPT-6.1 Terra: on an internal benchmark of error-inducing prompts, the error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra, and at maximum effort it fell to 4.3%. For software engineering, Terminal-Bench 4.0 results published by OpenAI show GPT-6.1 Terra solved 50.3% of tasks at maximum reasoning effort versus GPT-5.6 Terra’s published 21.5%.

OpenAI noted that for the most complex coding problems, GPT-6.1 Sol and GPT-6 Astra remain preferable. In Terminal-Bench-Science 0.1, OpenAI reported GPT-6.1 Terra surpassed GPT-6.1 Sol at lower cost and came within five points of Claude Opus 5.5 at under one-fifth the estimated cost.

On safety, OpenAI said GPT-6.1 Terra meets the Critical threshold under its Preparedness Framework and incorporates alignment improvements from the GPT-6 series, including stronger resistance to prompt injection and a higher likelihood of disclosing limitations.

Starting today, GPT-6.1 Terra is available to all Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex. Developers can access it through the OpenAI API as gpt-6.1-terra. Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens.


Original source: OpenAI News

Leave a Comment