Recommended Articles
Tech & AI

OpenAI Launches GPT-6.1 Terra, a Faster, Cheaper Model Built to Use Computers for You

By Angeled · October 27, 2026 · Tech & AI
OpenAI Launches GPT-6.1 Terra, a Faster, Cheaper Model Built to Use Computers for You
OpenAI Launches GPT-6.1 Terra, a Faster, Cheaper Model Built to Use Computers for You

OpenAI launched GPT-6.1 Terra on October 26, 2026, a new mid-tier model designed for the everyday work that fills an office day: reading documents, filling out forms, entering data, drafting proposals, and operating software the way a human user would. It is available to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and to developers through the API as gpt-6.1-terra. The headline is price as much as capability: at $1 per million input tokens and $5 per million output tokens, OpenAI says Terra costs 50% less than both its predecessor, GPT-5.6 Terra, and its larger sibling, GPT-6.1 Sol.

What Terra is for

Terra belongs to a class of AI often called "computer use" or agentic models. Unlike a conventional chatbot that only answers in text, a computer-use model can look at what is on a screen, click buttons, type into fields, and move between applications to complete a multi-step task. OpenAI positions Terra for visual reasoning and precise computer use, with clearer writing as a third strength. The example tasks it names are deliberately ordinary: drafting proposals, interpreting technical drawings, filling out forms, and editing videos. That ordinariness is the point. The most valuable AI work for most companies is not a brilliant answer to one hard question; it is thousands of routine tasks done reliably and cheaply.

Terra sits in the middle of OpenAI's GPT-6 family. Below it are the cheapest Luna models; above it are GPT-6.1 Sol and GPT-6 Astra, the most capable models, which OpenAI still recommends for the hardest coding and reasoning work. Terra replaces GPT-5.6 Terra and is aimed at high-volume work where the cost of each task decides whether automating it is worth doing at all.

Price, speed, and context

The API pricing is straightforward at standard rates: $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens. OpenAI says that is half the price of GPT-5.6 Terra. The model lists a context window of about 1.05 million tokens, with text and image input, meaning it can take in very long documents and screenshots in a single request. OpenAI also says Terra completes tasks more than three times faster on average than its larger Sol and Astra models, a claim that matters for agent workflows where a user is waiting for a multi-step job to finish.

On OpenAI's own benchmarks — not independent evaluations — Terra scored 60.0% on Agents' Last Exam at about $1.01 per task, and 50.3% on Terminal-Bench 4.0 at maximum effort, up from 21.5% for GPT-5.6 Terra. Founders should rerun their own tasks on Terra before switching production workloads.

Where it shipped on day one

Terra arrived the same day across several surfaces: the OpenAI API, ChatGPT Work and Codex, GitHub Copilot, and Amazon Bedrock, with Microsoft Foundry also carrying it. There is also a safety note founders should not skip. OpenAI says GPT-6.1 Terra meets the Critical threshold in cybersecurity under its Preparedness Framework, keeping the protections developed for GPT-6 Astra. In plain terms, OpenAI is saying the model is powerful enough in cybersecurity-related tasks that it needs the strongest safeguards in its framework. That is a statement about the model's own capability assessment, and it is a reminder that giving an agent the ability to click, type, and move data between applications is a security decision, not just a productivity one. Permissions, audit trails, and limits on what an agent can send or change deserve attention before deployment, not after.

Why computer use is the real story

The AI industry spent its first years competing on who could write the best answer. The current competition is about who can do the work: open the spreadsheet, reconcile the invoice, update the record, file the form. That shift changes the economics. A model that answers questions is priced per conversation; a model that does jobs is priced per completed task, and the winner is the model that finishes the most jobs at the lowest cost with the fewest errors. Terra's positioning — cheaper, faster, good enough for routine work — is aimed squarely at that market.

What it means for founders and investors

For founders, Terra is a prompt to reprice automation. If a workflow was shelved because agent costs were too high — invoice processing, data entry, lead enrichment, report drafting — it is worth testing again on a model priced at $1 and $5 per million tokens. Measure cost per completed task, not cost per token, and measure error rates on your own data. The cheapest model that finishes the job correctly is the right model, and Terra's pitch is that it is now that model for a large class of routine work.

For investors, Terra signals where the AI market is maturing. The frontier race continues, but the money in 2026 is increasingly in deployment economics: who can deliver capable agents at a price point where automation pays for itself. Companies building agent workflows, evaluation tooling, and permission and audit layers around computer-use models are riding that wave. The risk to watch is the same as the opportunity: agents that can act inside real software need trust, and trust is built slowly and lost quickly.

Angeled.com connects founders with investors.

Source: analyticsinsight.net
Analytics Insight, "OpenAI GPT-6.1 Terra Launches with App Control and Faster AI Workflows" (October 26, 2026)
Source: thepromptinsider.com
The Prompt Insider, "GPT-6.1 Terra: Price, Features, and Availability" (October 26, 2026)
Source: modeldex.substack.com
ModelDex, "OpenAI's GPT-6.1 Terra Halves the Price of AI That Can Use a Computer for You" (October 26, 2026)
Source: ainotis.com
AI Notis, "OpenAI says GPT-6.1 Terra meets Critical cybersecurity threshold, at half GPT-5.6 Terra's price" (October 26, 2026)

Read Next