OpenAI GPT-6.1 Terra: What's New, Price and Who Gets It

OpenAI has a new model aimed squarely at everyday office work. On Monday, Oct. 26, 2026, the company released GPT-6.1 Terra, which it describes as "a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing," in its announcement. It is cheaper than OpenAI's other current models and is available now in ChatGPT and the API. Searches for "OpenAI 6.1 Terra" passed 2,000 in the U.S. on Google Trends on Monday. Every performance figure below is OpenAI's own.

What we know about GPT-6.1 Terra

GPT-6.1 Terra is the successor to GPT-5.6 Terra and sits alongside OpenAI's more capable GPT-6.1 Sol and GPT-6 Astra, which OpenAI calls its most capable frontier model. OpenAI says Terra "excels at the communication and computer-based tasks that fill the workday, from drafting proposals and interpreting technical drawings to filling out forms and editing videos."

The Verge summed up the pitch: GPT-6.1 Terra "is very good at computer use, OpenAI says."

Sources: OpenAI; OpenAI model documentation.

Computer use: the headline feature

"Computer use" means an AI agent operating software the way a person would, by reading the screen and clicking and typing. OpenAI tested Terra on a benchmark it calls Agents' Last Exam, which includes tasks such as turning architectural drawings into a 3D model and reviewing an animated film shot by shot to create a time-coded log. According to OpenAI:

• At medium reasoning effort, Terra scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost.
• It performs on par with GPT-6 Astra at one-tenth Astra's price per token.
• At max effort, it is over three times as token-efficient as Anthropic's Claude Opus 5.5 at one-fourth its price per token, for what OpenAI estimates is a roughly 9x lower cost per task.
• It completes tasks over three times faster than GPT-6 Astra and GPT-6.1 Sol.

OpenAI also says Terra beat GPT-5.6 Terra at every reasoning setting on the MMMU-Pro visual reasoning benchmark, and that on OSWorld 2.1, a test of using desktop apps for multistep tasks, it surpassed GPT-5.6 Terra's best score at medium effort.

These are company-reported results. Independent testing will show how well they hold up.

Fewer errors and better coding

OpenAI says it has "invested heavily in reducing hallucinations." On its internal benchmark of error-inducing prompts:

• At low reasoning effort, the error rate fell from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra.
• At maximum effort, it fell by 18 points to 4.3%, nearly matching GPT-6 Astra and GPT-6.1 Sol.

For coding, OpenAI says Terra solves 50.3% of tasks on Terminal-Bench 4.0 at max effort, more than double GPT-5.6 Terra's published 21.5%. On Terminal-Bench-Science 0.1, it surpasses GPT-6.1 Sol at lower cost. But OpenAI is clear about the limits: "For the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice."

Pricing details for developers

OpenAI says Terra's standard API prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol. Its model documentation adds:

• Prompts with more than 272,000 input tokens are priced at 2x input and 1.5x output for the full request.
• Batch and Flex prices are 50% lower than Standard; Fast mode costs 2x.
• Reasoning effort can be set from none to max, with medium the default.
• Tool calling requires the Responses API, and supported tools include web search, file search, code interpreter and computer use.

For ChatGPT subscribers, 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.

Safety

OpenAI says GPT-6.1 Terra meets the Critical threshold in cybersecurity under its Preparedness Framework and keeps the protections developed for GPT-6 Astra. It also says Terra is more likely than GPT-5.6 Terra to stay within scope and disclose its limitations, and showed strong resistance in prompt-injection tests. Details are in OpenAI's system card addendum.

What the system card adds

The addendum fills in more detail. OpenAI says it is treating Terra as Critical capability in cybersecurity, High in the biological and chemical domain and below the High threshold in AI self-improvement, and applies the same safeguards as GPT-6.1 Sol and GPT-6 Astra. Among the reported results:

• On MentalHealthBench, a benchmark built with mental health experts, Terra scored 55.5%, up from 46.2% for GPT-5.6 Terra.
• In a simulated deployment across 48,690 internal Codex tasks, Terra received 22 high-severity misalignment flags (0.045%), fewer than the older GPT-6 Sol's 38 and GPT-5.6 Sol's 61.
• In a test of how models react when an automated reviewer blocks an action, Terra made no attempts to bypass the monitor.
• Not every number improved: in a "respecting warnings" test, Terra completed a task through a route that violated the spirit of a restriction in 29.6% of rollouts at maximum effort, higher than GPT-6 Astra and GPT-6.1 Sol. OpenAI notes the test covers mostly low-stakes situations and is run without system-level controls.

What's next

1. Independent benchmarks. Expect third-party testing of OpenAI's speed, cost and accuracy claims in the coming weeks.
2. Developer adoption. OpenAI is pitching Terra for high-volume automation, such as reading invoices and entering details into accounting software.
3. More tech news. Apple also had a busy Monday; see the new OLED iPad mini. For AI and markets, read about Anthropic's IPO filing.

Who Terra is for

Terra is the model OpenAI wants doing the high-volume office work, like operating apps, reading drawings and filling out forms, at $1 and $5 per million tokens. It isn't meant for the hardest coding jobs; OpenAI still points those to Sol and Astra. And until outside testers weigh in, the speed and accuracy gains are OpenAI's word.