GPT-6.1 Terra: Overview
GPT-6.1 Terra (gpt-6.1-terra) is OpenAI's fast, cost-efficient model for visual reasoning, computer-use workflows, and clear natural-language writing. It is designed for high-volume professional tasks where a model must interpret screenshots or images, interact with software, and produce useful written output without the cost profile of a flagship model. OpenAI announced it on October 26, 2026.
Technical specifications of GPT-6.1 Terra
| Specification | Publicly documented details |
|---|---|
| Provider / family | OpenAI; GPT-6 series |
| API model ID | gpt-6.1-terra |
| Input modalities | Text and images |
| Output modality | Text |
| Context window | 1,050,000 tokens |
| Maximum output | 128,000 tokens |
| Knowledge cutoff | June 1, 2026 |
| Reasoning effort | none, low, medium (default), high, xhigh, max |
| Tool use | Function calling and Responses API tools including web search, file search, image generation, code interpreter, hosted shell, apply patch, skills, computer use, MCP, and tool search |
| Streaming / structured outputs | Supported |
| API endpoints listed | Responses, Chat Completions, Batch, and other platform endpoints; tool calling requires the Responses API |
Source: OpenAI API model documentation. Endpoint availability and features can differ by product; check the current model page before implementation.
What is GPT-6.1 Terra?
GPT-6.1 Terra is a model tuned for practical work that combines visual understanding with computer-use capability and concise, natural writing. Representative tasks include reading information from invoices and entering it into accounting software, interpreting technical drawings, filling out forms, reviewing visual material, drafting proposals, and making scoped code changes. It is positioned for throughput-sensitive workflows rather than as OpenAI's strongest option for the most difficult reasoning or software-engineering projects.
Main features of GPT-6.1 Terra
- Visual reasoning: Accepts image input and is designed to interpret screenshots, technical drawings, and other visual material.
- Computer use: Can be paired with the Responses API's computer-use tools to operate across software interfaces in multi-step workflows.
- Large context: A documented 1.05-million-token context window supports long instructions and substantial reference material.
- Configurable reasoning:
reasoning.effortranges fromnonethroughmax, allowing developers to trade speed/cost against additional deliberation. - Tool-enabled workflows: Supports function calling and a broad set of Responses API tools, including web search, file search, code interpreter, hosted shell, and MCP.
- Efficient professional writing and coding: OpenAI positions Terra for natural writing, repetitive workday automation, and scoped software-engineering tasks at a lower standard token price than GPT-6.1 Sol or GPT-6 Astra.
Benchmark Performance of GPT-6.1 Terra
OpenAI reports improvements across computer use, visual reasoning, factuality, and coding. The following results are published in its launch announcement.
OpenAI
| Benchmark / evaluation | Reported result | Interpretation |
|---|---|---|
| Agents’ Last Exam | 15 percentage points above GPT-5.6 Terra at medium reasoning effort | Stronger performance on professional computer-use tasks |
| Internal error-inducing prompts | Error rate of 16.7% at low effort; 4.3% at maximum effort | Fewer factual errors on this specific internal evaluation |
| Terminal-Bench 4.0 | 50.3% at maximum reasoning effort | Improved command-line coding and software engineering |
| MMMU-Pro | Outperforms GPT-5.6 Terra at every tested reasoning setting | Improved visual reasoning |
| OSWorld 2.1 | At medium effort, exceeds GPT-5.6 Terra’s best score | Improved multistep desktop-app operation |
Benchmark caveat: These results use different evaluation protocols and reasoning settings. They should not be treated as directly comparable percentages of overall model quality. OpenAI's announcement does not provide every underlying numerical score in its text, so those values are intentionally omitted.
GPT-6.1 Terra vs GPT-6.1 Sol vs GPT-6 Astra
| Model | Primary strength | Best-fit workload |
|---|---|---|
| GPT-6.1 Terra | Visual reasoning, computer use, speed and cost efficiency | Document processing, screen-based automation, routine coding |
| GPT-6.1 Sol | More demanding, sustained reasoning and complex workflows | Long-running coding and multi-step projects |
| GPT-6 Astra | Highest-end complex reasoning and judgment within this comparison | Difficult end-to-end research, coding and agentic tasks |
| Claude Opus 5.5 | Alternative for advanced reasoning and agentic work | Complex tasks requiring careful evaluation |
OpenAI positions Terra as a more economical option for everyday professional work, while recommending GPT-6.1 Sol and GPT-6 Astra for the most complex coding tasks.
Limitations of GPT-6.1 Terra
- No direct audio or video input: The model documentation lists text and images as input modalities. Video and audio processing may require separate preprocessing or specialized endpoints.
- Computer use requires integration: Application automation depends on supported tools, the Responses API, execution environments and appropriate permissions.
- Reasoning has a cost–latency trade-off: Higher reasoning effort can improve performance on difficult tasks but may increase latency and token usage.
- Not the default choice for the hardest coding tasks: OpenAI recommends GPT-6.1 Sol or GPT-6 Astra for its most demanding coding workflows.
- Benchmark results are task-specific: Performance on internal or public evaluations does not guarantee equivalent results in every production environment.
Recommended Use Cases
- Invoice and document automation: Extract fields from incoming invoices and populate accounting systems.
- Visual quality assurance: Inspect screenshots, charts, product images and technical drawings.
- Business application agents: Fill forms, navigate supported interfaces and execute repetitive workflows.
- Professional writing: Draft proposals, revise reports and transform source materials into structured documents.
- Routine software engineering: Handle focused coding, debugging and command-line tasks where cost efficiency matters.
How to Access the GPT-6.1 Terra API Through CometAPI
No coding required for account setup and initial testing.
Step 1: Sign Up for CometAPI
Visit CometAPI and create an account, or sign in if you already have one. Open the dashboard to access the platform's available AI models and API services.
Step 2: Get Your API Key
Navigate to the API token or key management section in your CometAPI console. Create a new API key and store it securely. This key authenticates your requests to the platform, so never share it publicly.
Step 3: Find GPT-6.1 Terra
Open the model directory and search for GPT-6.1 Terra. If available, select the model and confirm its exact API model ID is gpt-6.1-terra. Review its supported endpoints and current usage requirements before proceeding.
GPT-6.1 Terra is designed for visual reasoning, computer use, and natural writing.
Step 4: Test the API Without Writing Code
CometAPI provides a Playground or online API testing interface for the model:
- Select
gpt-6.1-terra. - Enter a simple prompt, such as “Summarize the benefits of AI automation in three bullet points.”
- Submit the request and review the response.
- If the test succeeds, consult the model's API documentation to connect it to your application or workflow.