TL;DR
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
OpenAI announced GPT-6.1 Terra, a model for visual reasoning, computer use, writing and coding. The company reports gains over GPT-5.6 Terra on its benchmarks and says standard API prices are 50% lower than for GPT-5.6 Terra and GPT-6.1 Sol. It is available through the API and to eligible ChatGPT Work and Codex subscribers.
OpenAI has introduced GPT-6.1 Terra, a model designed for visual reasoning, computer use, writing and coding, and says it is available through the API and to eligible ChatGPT Work and Codex users. Learn more about whether Terra is right for your AI workload. The company says the model improves on GPT-5.6 Terra in several tests while standard API prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol, a pricing change aimed at making repeated workplace automation less costly.
OpenAI describes Terra as a fast, efficient model for tasks such as interpreting technical drawings, filling out forms, drafting proposals and editing videos. Developers can access it through the API under the name gpt-6.1-terra. OpenAI lists standard prices of $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. Those prices are the company’s published rates; actual costs depend on token use and the work being automated.
The company reports gains on multiple benchmarks, including reported GPT-6.1 benchmark results. On Agents’ Last Exam, which tests professional tasks involving computer use, Terra scored 15 percentage points above GPT-5.6 Terra at medium reasoning effort, according to OpenAI, at about one-third the cost. OpenAI also says Terra completed tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol in its testing. On Terminal-Bench 4.0, a command-line coding benchmark, the model solved 50.3% of tasks at maximum reasoning effort, compared with GPT-5.6 Terra’s published score of 21.5%.
OpenAI also reports lower error rates on an internal set of prompts designed to elicit mistakes: 16.7% for Terra at low reasoning effort, against 53.6% for GPT-5.6 Terra; at maximum effort, Terra’s rate was 4.3%. These are company-reported results, not independent evaluations. OpenAI says GPT-6.1 Sol and GPT-6 Astra remain better choices for its most complex coding tasks; see this introduction to GPT-6.1 Sol.
Lower Costs for Workplace Automation
Terra’s stated price and performance combination could matter most to organizations that process large volumes of routine computer-based work. OpenAI gives the example of reading incoming invoices and entering their details into accounting software. If the reported speed and accuracy hold up in real workflows, lower per-token prices could make it more practical to run such tasks repeatedly, though token pricing alone does not establish the total cost or reliability of an automated process.
The model also sits below OpenAI’s more capable offerings in some use cases. The company says it approaches the performance of GPT-6 Astra on selected visual tests at a lower token price, while recommending its other models for the most complex coding work. That positions Terra as a lower-cost option for selected workday tasks, rather than a replacement for every model or a guarantee that a task can run without human review.
For users, availability through ChatGPT Work and Codex means access is not limited to developers building API integrations. OpenAI says the same usage allowance covers more tasks with Terra than with GPT-5.6 Terra, and substantially more than with GPT-6.1 Sol or GPT-6 Astra. The company has not provided a universal measure of how many additional tasks a user can complete, so the practical difference will depend on the work and usage pattern.
Products Worth Considering
Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Bose QuietComfort Wireless Noise Cancelling Headphones, Twilight Blue
As an affiliate, we earn on qualifying purchases.
Terra in OpenAI’s Model Lineup
GPT-6.1 Terra follows GPT-5.6 Terra and is part of OpenAI’s GPT-6.1 model family. The company presents it as a model focused on communication and computer-based tasks, including work that combines visual interpretation with actions in desktop software. Its announcement compares Terra with GPT-5.6 Terra, GPT-6.1 Sol, GPT-6 Astra and, on some tests, Anthropic’s Claude Opus 5.5.
The benchmarks named in the announcement test different abilities and should not be treated as interchangeable. Agents’ Last Exam and OSWorld 2.1 assess computer-use tasks; MMMU-Pro evaluates visual reasoning; Terminal-Bench 4.0 tests command-line coding; and Terminal-Bench-Science 0.1 covers coding for scientific research tasks. OpenAI says Terra outperforms GPT-5.6 Terra on MMMU-Pro at every reasoning setting, but benchmark performance does not by itself show how a model will perform across every workplace application.
Independent Results and Real-World Costs
The performance and safety findings in the announcement come from OpenAI’s own evaluations. The source does not provide independent validation of the benchmark results, nor enough information to determine how representative the tested tasks are of typical business use. OpenAI refers readers to a system card for more detail on safety evaluations, but the announcement does not reproduce the full methodology or results.
It is also unclear how Terra will perform across specific third-party applications, or how often human intervention will be needed when it handles documents, forms or desktop software. API rates are stated per million tokens; total spending will vary with input and output volume, repeated attempts and the design of an application. OpenAI’s claim that users can complete more tasks within the same plan allowance is not accompanied here by a numerical task estimate.
The announcement says Terra meets the “Critical” threshold in cybersecurity under OpenAI’s Preparedness Framework and retains protections developed for GPT-6 Astra. That is the company’s characterization of its internal assessment. The source does not provide enough information to independently assess the threshold or the model’s behavior in every deployment setting.
Availability and User Testing
OpenAI says GPT-6.1 Terra is available starting today to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, as well as to developers through the API. The company’s announcement does not give a calendar date, so the precise launch date cannot be established from the supplied material.
Users and developers can now test the model on their own tasks and compare its quality, speed and total costs with alternatives. Further evidence would include independent benchmark results, detailed deployment data and information about how reliably Terra completes multistep tasks without human correction. OpenAI has not announced in the supplied material a date for additional Terra releases or a schedule for further evaluation disclosures.
Key Questions
What is GPT-6.1 Terra?
It is an OpenAI model designed for visual reasoning, computer use, writing and coding, including tasks that involve interacting with desktop applications.
How much does GPT-6.1 Terra cost through the API?
OpenAI lists standard rates of $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. The company says standard prices are 50% below those for GPT-5.6 Terra and GPT-6.1 Sol.
Where is GPT-6.1 Terra available?
OpenAI says it is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers through the API as gpt-6.1-terra.
Are the reported benchmark results independently verified?
The supplied announcement presents results from OpenAI’s evaluations. It does not include independent verification, so the results should be understood as company-reported findings.
Is Terra OpenAI’s best model for every coding task?
No. OpenAI says GPT-6.1 Sol and GPT-6 Astra remain better choices for its most complex coding tasks.
Source: openai.com
