# GPT-6.1 Terra: OpenAI's half-price agentic model, explained

> The Oct. 26 release pairs a 1.05-million-token context window and computer use with API pricing that is 50% below GPT-6.1 Sol.

*Published 2026-10-27 · By Marcus Vance*

GPT-6.1 Terra is OpenAI's Oct. 26, 2026, model release that pairs a 1,050,000-token context window and up to 128,000 output tokens with visual reasoning, computer use, structured outputs, function calling and web search at $1 per million input tokens and $5 per million output tokens, half the price of GPT-6.1 Sol.

OpenAI released GPT-6.1 Terra on Oct. 26, 2026, adding a lower-cost, agent-focused model to the GPT-6.1 family. The company's announcement describes the release as a fast, efficient model that combines visual reasoning and computer use with clearer, more natural writing. Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens, according to OpenAI. That pricing is 50% lower than GPT-6.1 Sol and GPT-5.6 Terra, OpenAI said.

GPT-6.1 Terra is available through the OpenAI API under the model ID gpt-6.1-terra and is generally available in GitHub Copilot, per GitHub's changelog. The model documentation lists text and image input modalities, a 1,050,000-token context window and up to 128,000 max output tokens. Supported features include streaming, structured outputs, function calling, web search and computer use.

The release date, model ID and pricing make Terra a concrete option for teams that were waiting on a cheaper GPT-6.1-class model. The documentation does not publish latency figures, so speed claims in the announcement are not yet independently benchmarked.

## What is GPT-6.1 Terra and how is it positioned?

OpenAI has positioned Terra as a fast, efficient complement to GPT-6.1 Sol rather than a replacement. The API price is half that of Sol, and GitHub's changelog describes Terra as designed for everyday agentic coding and terminal work at a lower cost than Sol. The naming convention, with Terra alongside Sol, gives developers two points on a cost-versus-capability curve within the same GPT-6.1 generation.

> We’re introducing GPT-6.1 Terra, a fast, efficient model that combines detailed visual reasoning and precise computer use alongside clearer, more natural writing.OpenAI, company announcement

## What does GPT-6.1 Terra add to OpenAI's API?

Terra's 1,050,000-token context window allows a single request to hold large codebases, long terminal sessions or multimodal documents. The 128,000-token output limit supports long generated artifacts, including full files and multi-step plans. Those limits come from OpenAI's model documentation for gpt-6.1-terra.

The model's supported feature set is aimed at agentic products. Computer use lets a model act on graphical interfaces, while function calling and structured outputs keep results machine-readable. Web search is listed as a supported feature in the model documentation. OpenAI did not release per-feature performance benchmarks for Terra in the reviewed materials.

| Specification | GPT-6.1 Terra |
| --- | --- |
| Context window | 1,050,000 tokens |
| Max output tokens | Up to 128,000 |
| Input modalities | Text, image |
| Input price | $1 per million tokens |
| Cached input price | $0.05 per million tokens |
| Output price | $5 per million tokens |
| Model ID | gpt-6.1-terra |
| Supported features | Streaming, structured outputs, function calling, web search, computer use |
| GitHub Copilot | Generally available |

The table summarizes the specifications OpenAI published for the model. The combination of a large context window, image input and computer use is what makes Terra usable for browser agents and terminal-based coding agents.

- Call the model directly through the OpenAI API using the model ID gpt-6.1-terra.
- Run workloads with text and image inputs in a context window of up to 1,050,000 tokens.
- Generate up to 128,000 output tokens per request for long-form reasoning and code generation.
- Enable structured outputs, function calling, web search and computer use through supported API features.
- Use the model inside GitHub Copilot for everyday agentic coding and terminal work, per GitHub's changelog.

## How much does GPT-6.1 Terra cost compared with GPT-6.1 Sol?

OpenAI's announcement states that standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. The same announcement says the pricing is 50% lower than GPT-6.1 Sol and GPT-5.6 Terra. For high-volume workloads, the cached-input price matters because agent loops frequently resubmit the same system prompts and context.

A 50% reduction in both input and output price changes the breakeven point for agent deployments that previously had to use smaller models. It also makes retries cheaper, which is relevant because agentic systems often fail and retry. GitHub's changelog ties the lower cost to everyday coding and terminal work.

## How does GPT-6.1 Terra perform on agentic benchmarks?

OpenAI reported two benchmark results for Terra. On Terminal-Bench 4.0 at maximum reasoning effort, Terra solved 50.3% of tasks, compared with 21.5% for GPT-5.6 Terra. The same announcement reports an error rate of 16.7% on an internal benchmark of error-inducing prompts at low reasoning effort, down from 53.6% for GPT-5.6 Terra.

The Terminal-Bench result is reported at maximum reasoning effort, so it describes the model's ceiling on terminal-agent tasks rather than its behavior under latency constraints. The error-rate result is measured at low reasoning effort, which is closer to how a cost-conscious deployment would run the model. OpenAI did not publish a breakdown of how reasoning-effort settings affect the error rate in the reviewed materials.

## What do the benchmark numbers mean for computer-use and visual reasoning workloads?

Computer-use agents depend on the model reading screenshots and acting on UI elements. Terra's image input support and computer-use feature are what connect visual reasoning to action. The error-rate improvement at low reasoning effort is particularly relevant for these workloads because UI agents send many small, repeated requests.

OpenAI's announcement says Terra combines detailed visual reasoning and precise computer use alongside clearer, more natural writing. The model documentation confirms image inputs and computer use as supported capabilities. The benchmark evidence, however, is limited to terminal tasks and error-inducing prompts; OpenAI did not release separate computer-use benchmark scores in the reviewed materials.

## What is the role of GitHub Copilot in GPT-6.1 Terra's rollout?

GitHub announced the model's general availability in Copilot on the same day as OpenAI's release. The changelog says GPT-6.1 Terra is designed for everyday agentic coding and terminal work at a lower cost than GPT-6.1 Sol. The integration gives developers a path to Terra without building a separate API pipeline.

> GPT-6.1 Terra, OpenAI’s newest model, is now generally available in GitHub Copilot. It is designed for everyday agentic coding and terminal work at a lower cost than GPT-6.1 Sol.GitHub, GitHub Changelog

Copilot's integration targets the terminal as well as code completion. Terminal work is where agentic loops run commands, parse output and correct errors, so a model with lower error rates can reduce the number of command cycles. GitHub's changelog does not specify which Copilot plans include Terra, so pricing details for the integration are not available from the reviewed source.

## What are the enterprise implications of a half-price GPT-6.1 model?

Enterprises running agentic systems face two cost drivers: tokens and retries. Terra lowers the per-token price while improving the low-reasoning error rate, which should reduce both drivers. OpenAI's pricing announcement and its internal error benchmark are the evidence for that claim.

The release also creates a pricing anchor for mid-tier agent models. Any competitor selling computer-use or agentic-coding models above $1 input and $5 output now has to justify the premium. OpenAI's model documentation makes the capabilities and price transparent to procurement teams.

## What do OpenAI and GitHub say about GPT-6.1 Terra?

OpenAI's announcement describes Terra as a fast, efficient model that combines detailed visual reasoning and precise computer use with clearer, more natural writing. GitHub's changelog frames the same model as an everyday agentic coding and terminal work option at a lower cost than Sol. Together, the two sources define Terra's role: a cheaper agentic workhorse with a large context window.

## What comes next after GPT-6.1 Terra?

The reviewed OpenAI and GitHub materials do not include a roadmap for additional GPT-6.1 variants. What is known is that Terra entered the API and Copilot on Oct. 26, 2026, with a stable model ID and published pricing. Teams can migrate workloads from Sol to Terra where cost is the binding constraint, while keeping Sol for tasks that need its higher price tier.

Developers should watch for updated model documentation, new benchmark results and pricing changes, since OpenAI's announcement and model page are the sources of record for Terra's specifications. The next natural checkpoint is evidence of how Terra performs in production computer-use and terminal workloads, beyond the two benchmarks in the launch announcement.

GPT-6.1 Terra is not a frontier capability jump in the sense of a larger context window alone; its significance is the combination of that window, agentic tooling and a 50% price cut. For AI teams, the release makes high-volume agentic workflows materially cheaper on paper, with the documented error-rate reduction supporting the business case.

## Sources

1. [GPT-6.1 Terra launched Oct. 26, 2026, with standard API pricing of $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens; on Terminal-Bench 4.0 at maximum reasoning effort it solved 50.3% of tasks versus 21.5% for GPT-5.6 Terra, and its low-reasoning error rate dropped to 16.7% from 53.6%.](https://openai.com/index/introducing-gpt-6-1-terra/)
2. [The model ID gpt-6.1-terra supports text and image inputs, streaming, structured outputs, function calling, web search, computer use, a 1,050,000-token context window and up to 128,000 max output tokens.](https://developers.openai.com/api/docs/models/gpt-6.1-terra.md)
3. [GPT-6.1 Terra is generally available in GitHub Copilot and is designed for everyday agentic coding and terminal work at a lower cost than GPT-6.1 Sol.](https://github.blog/changelog/2026-10-26-gpt-6-1-terra-in-github-copilot/)

---
Source: https://aiintelreport.com/frontier-models/gpt-6-1-terra-openai-half-price-agentic-model
Index: https://aiintelreport.com/llms.txt · Full text: https://aiintelreport.com/llms-full.txt