If you have been following OpenAI’s rapid model releases, you may be wondering: what is GPT-6.1 Terra, and should you actually use it? In short, GPT-6.1 Terra is OpenAI’s fast, efficient model built for everyday work that involves reading screens, understanding visuals, and getting things done on a computer — from drafting proposals to entering invoices into accounting software.
OpenAI launched GPT-6.1 Terra on October 26, 2026, for ChatGPT Work and Codex (Plus, Pro, Business, Enterprise, and Edu plans) and through the API as gpt-6.1-terra. One thing to be clear about up front: Terra is not available in regular ChatGPT chat. It lives in the Work and Codex environments, where computer use and coding workflows are the main focus.
This guide covers what Terra does well, what it costs, how it compares, and how to select it in Codex.
What Is GPT-6.1 Terra and What Is It Built For?
GPT-6.1 Terra is the efficient workhorse of the GPT-6.1 lineup: fast, with detailed visual reasoning, precise computer use, and clearer, more natural writing than its predecessor. That makes it a strong fit for everyday office work, including:
- Drafting proposals, reports, and client documents
- Reading technical drawings and extracting the details that matter
- Filling out forms accurately, without skipping fields
- Editing videos as part of a guided workflow
- Entering invoices into accounting software
- Handling routine computer tasks that require both vision and action
Think of Terra less as a model you ask a single question, and more as a model you hand a job to — one that involves looking at something on screen, understanding it, and doing the next step correctly. Terra replaces GPT-5.6 Terra. It sits above Luna and below Sol and Astra: good enough for most daily work, at a price that makes heavy use affordable.
GPT-6.1 Terra Pricing: What It Actually Costs
One of the biggest reasons Terra stands out is its price. GPT-6.1 Terra API standard pricing is:
- $1 per million input tokens
- $0.05 per million cached input tokens
- $5 per million output tokens
There are also discounted and premium options to know about:
- Batch and Flex: 50% less than standard pricing, ideal for work that does not need an instant response.
- Fast mode: 2x standard pricing, for when speed matters most.
- Very long prompts: For prompts over 272,000 input tokens, input and cached input rates double, and output rises by 50% for the full request.
Compared with Sol, Terra costs half as much, and compared with Astra, the gap is even wider. For high-volume work, that difference adds up fast.
GPT-6.1 Terra vs Sol: Which One Should You Pick?
- Choose Terra for everyday office work, visual reasoning, form filling, document drafting, and precise computer use at scale.
- Choose Sol for the most complex coding tasks, where OpenAI itself says Sol and Astra remain better.
Terra is not trying to beat Sol at everything. It aims to win on speed, efficiency, and cost for everyday work, while Sol handles the hardest engineering problems.
GPT-6.1 Terra vs Luna, Sol, and Astra: Comparison Table
| Model | Input price (per million tokens) | Output price (per million tokens) | Position in the family |
|---|---|---|---|
| Luna | Lower than Terra (OpenAI has not published a separate figure in this guide) | Lower than Terra | Smallest, most affordable tier |
| GPT-6.1 Terra | $1 | $5 | Fast, efficient, everyday work |
| Sol | $2 | $10 | More powerful, complex coding |
| Astra | $10 | $50 | Top tier, most capable |
Terra costs half as much as Sol and one-tenth as much as Astra on input.
GPT-6.1 Terra Specifications at a Glance
- Context window: 1,050,000 tokens
- Maximum input: 922,000 tokens
- Maximum output: 128,000 tokens
- Knowledge cutoff: June 1, 2026
- Inputs accepted: Text and images
- Outputs: Text
- Tool calling: Requires the Responses API
- Codex context: Codex lists a separate 500,000-token context for Terra
The huge context window means Terra can work with long documents and large codebases without losing track of earlier details, though the Codex context is smaller than the full API window. With a June 1, 2026 cutoff, Terra will not know about later events unless you provide the information or connect it to live tools.
How to Use GPT-6.1 Terra in Codex
Switching to Terra is straightforward. You need Codex CLI version 0.159.0 or later. Select Terra in any of these ways:
- From the command line: run
codex -m gpt-6.1-terra - Inside a session: type
/modeland choose GPT-6.1 Terra - In your configuration file: set
model = "gpt-6.1-terra"inconfig.toml
In ChatGPT Work or the desktop app, pick Terra from the model list under Advanced. For the API, use gpt-6.1-terra; tool calling requires the Responses API.
GPT-6.1 Terra Computer Use: What Makes It Different?
First, detailed visual reasoning means Terra can look at an image, a drawing, a screenshot, or a scanned form and understand its structure — not just read the text, but grasp what each part means. Second, precise computer use means it can then act on that understanding accurately, whether that means filling a field, selecting the right option, or completing a multi-step task without drifting off course.
That combination is why Terra fits invoice entry and form completion so well: those jobs reward care, consistency, and speed across hundreds of repetitions. OpenAI’s own vendor benchmarks — its claims, not independently verified results — point in that direction:
- Terminal-Bench 4.0: 50.3% at maximum effort, compared with 21.5% for GPT-5.6 Terra
- Factual-error rate: fell from 53.6% to 16.7% at low effort
- Difficult-prompts evaluation: 4.3% at maximum effort
Those numbers suggest a meaningful jump over GPT-5.6 Terra, but treat vendor benchmarks as a starting point, not proof — the real test is your own documents and workflows.
GPT-6.1 Terra API Price Per Million Tokens: Is It Worth It?
For most everyday workloads, yes — provided Terra matches your task type. At $1 per million input tokens and $5 per million output tokens, with cached input at just $0.05, Terra is built for volume, and Batch and Flex pricing makes non-urgent jobs even cheaper. For the most complex coding work, though, OpenAI is clear that Sol and Astra remain better. A practical rule of thumb:
- High volume + routine or visual work = Terra
- Low volume + extremely complex code = Sol or Astra
- Unsure? Start with Terra and upgrade only where errors cost more than the price difference.
If you need help choosing AI tools, explore our services, or contact us. For more AI tool guides, see our Ideogram Vector SVG guide and free AI tools for small business marketing.
Frequently Asked Questions About GPT-6.1 Terra
Is GPT-6.1 Terra available in regular ChatGPT?
No. It launched for ChatGPT Work and Codex (Plus, Pro, Business, Enterprise, and Edu plans) and through the API. It is not in regular ChatGPT chat.
How much does GPT-6.1 Terra cost?
Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens. Batch and Flex are 50% less, Fast mode is 2x, and prompts over 272,000 input tokens carry higher rates, as detailed above.
GPT-6.1 Terra vs Sol: which should I pick?
Pick Terra for everyday office work, visual reasoning, and precise computer use at scale. Pick Sol for the most complex coding tasks, where OpenAI says Sol and Astra remain stronger.
How do I select GPT-6.1 Terra in Codex?
Use Codex CLI version 0.159.0 or later, then run codex -m gpt-6.1-terra, use /model inside a session, or set model = "gpt-6.1-terra" in your config.toml. In the Work or desktop app, choose it from the model list under Advanced.
What is GPT-6.1 Terra’s knowledge cutoff?
June 1, 2026. Terra does not know about events after that date unless you provide current information or connect it to live data sources.
Conclusion: Should You Try GPT-6.1 Terra?
So, what is GPT-6.1 Terra in one sentence? It is OpenAI’s fast, affordable, computer-use model for everyday work — a step up from GPT-5.6 Terra, cheaper than Sol and Astra, and built for the visual, routine, and screen-based tasks that fill a normal workday.
If you work in ChatGPT Work or Codex, or you build with the API, Terra is worth testing on your real workflows, especially if cost and volume matter. Just remember its limits: it is not in regular ChatGPT, its knowledge stops at June 1, 2026, and for the most complex coding, Sol and Astra are still stronger.
Sources: Skimwire, Nerdschalk, Aistify.
