GPT-6.1 Terra is OpenAI’s fast, cost-conscious model for visual reasoning, computer use, and everyday professional work. It is designed for tasks such as reading documents, interpreting screenshots and technical drawings, filling out forms, and operating software through supported tools. This profile summarizes the provider’s published specifications and pricing as of October 26, 2026; check the official documentation before deploying because model details and rates can change.
GPT-6.1 Terra at a glance
| Specification | Published details |
|---|---|
| API model ID | gpt-6.1-terra |
| Announced | October 26, 2026 |
| Context window | 1,050,000 tokens |
| Maximum output | 128,000 tokens |
| Input and output | Text and image input; text output |
| Reasoning effort | None, low, medium, high, xhigh, and max |
| Tool calling | Use the Responses API for tool calling |
| Fine-tuning | Not supported in the published model specification |
Sources: OpenAI’s GPT-6.1 Terra model documentation and the official announcement.
API pricing
OpenAI lists these standard rates in US dollars per million tokens (MTok):
| Token type | Standard price per 1M tokens |
|---|---|
| Input | $1.00 |
| Cached input | $0.05 |
| Output | $5.00 |
| Cache write, 30-minute retention | $1.25 |
| Cache write, 12-hour retention | $2.50 |
Batch and Flex processing are listed at 50% below Standard rates. Fast mode costs twice the applicable Standard rate, and regional processing may add a 10% premium where available. Prompts with more than 272,000 input tokens have different long-context multipliers, so do not estimate every request using the headline rates alone. Tool-specific fees may also apply. See the official API pricing page for current terms.
What is GPT-6.1 Terra best suited for?
- Computer-use workflows: interpret a screen, locate controls, and perform bounded multi-step tasks with suitable permissions and verification.
- Document and image understanding: extract information from invoices, screenshots, product photos, and technical drawings.
- High-volume office automation: draft, classify, summarize, or transform routine work where latency and cost matter.
- Structured extraction: convert visual or text inputs into fields that downstream software can validate.
- Assisted coding and debugging: handle practical engineering tasks, while more complex coding work may justify testing GPT-6.1 Sol or GPT-6 Astra as alternatives.
These are use-case suggestions, not guarantees. OpenAI’s published benchmark results are vendor-reported; test the model on your own workload before relying on those results for a purchasing decision.
Important limitations and trade-offs
- Images are input, not a native output modality: the model specification lists image input and text output. Image generation is available as a supported tool, but audio and video are not listed as supported direct modalities.
- Tool calling requires the Responses API: review the endpoint requirements before changing an existing integration.
- Long context can raise the bill: the published pricing applies higher multipliers to requests above 272K input tokens.
- Benchmarks do not replace task testing: evaluate accuracy, latency, tool-call reliability, completion rate, and human correction effort.
- Computer use needs safeguards: restrict permissions, isolate sensitive actions, confirm irreversible operations, and log what the agent does.
How to evaluate GPT-6.1 Terra
- Choose 30–100 representative tasks, including edge cases and failures.
- Record success rate, task completion time, token usage, retries, and any separate tool charges.
- Compare the same tasks against a baseline model using identical instructions and acceptance criteria.
- Calculate cost per successful task, including retries and human review—not only token cost per request.
- Start with low-risk, read-only actions; add write or submit permissions only after testing and adding approval gates.
GPT-6.1 Terra vs. GPT-6.1 Sol
Terra is positioned as a faster, lower-cost option for visual understanding and computer-use work. Sol is a stronger alternative to evaluate when a workflow needs more complex reasoning or coding performance. Neither choice should be made from price alone: compare completion quality and cost on the tasks that matter to your team.
Read our GPT-6.1 Sol profile and the Gemini 4 Argon vs. GPT-6.1 Sol comparison. For broader computer-use options, see AI computer-use tools for 2026, or browse the Aiverio AI tools directory.
Frequently asked questions
What is the API model ID?
The published ID is gpt-6.1-terra. Confirm the current model catalog before putting the ID into production code.
Does GPT-6.1 Terra support computer use?
OpenAI lists computer use among the tools supported through the Responses API. Real-world reliability depends on the interface, permissions, instructions, and verification steps in your implementation.
Is it cheaper than GPT-6.1 Sol?
OpenAI’s published standard input and output token prices for Terra are half the corresponding standard rates for GPT-6.1 Sol. Actual task cost depends on reasoning effort, token volume, caching, processing tier, tools, and retries.
