OpenAI just shipped GPT-6.1 Terra, a model that can see a screen, reason about what it shows, and take action on it—filling forms, processing invoices, editing videos—at API prices 50% lower than its predecessor. It is available today to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and to developers through the API.
What it takes on
The tasks Terra is built for are not novel; they are the ones that eat afternoons. Reading an incoming invoice and entering its details into accounting software. Converting a technical drawing into a 3D model. Watching an animated film shot by shot and producing a time-coded restoration log. What is new is the cost. On OpenAI's internal factuality benchmark, the error rate at low reasoning effort fell from 53.6% with the previous Terra model to 16.7% with this one. That is cheap enough to run on every transaction rather than a careful sample.
The work that stays with a person is the exception: the invoice that doesn't parse, the drawing with an ambiguity, the editorial call. Terra handles the queue; a human handles the edge. Operations teams running high-volume document and media workflows gain the most. The team exposed is anyone selling manual data-entry services or single-purpose software that does one of these jobs without an agentic layer underneath it.
The trend it fits
This is the same bet Anthropic is making with tiered Claude pricing and Google is making with Gemini Flash: capability is table stakes, cost per task is the battleground. The vendors that find the point where a model is accurate enough to run autonomously and cheap enough to run on every transaction win the infrastructure contract. Terra is OpenAI's claim on that point for visual and computer-use work.
What matters at scale is not the benchmark number but what the economics permit. When agents cost a fraction of what they did on the previous Terra model, the question stops being whether to automate a workflow and starts being which workflows to leave manual. Finance teams, media operations, and any back-office function that moves documents through a system are looking at a different cost structure than they had. PE partners underwriting businesses built on manual processing volume should be asking the same question their portfolio companies are about to face.
Get the Daily Frontier
Stories like this one land in your inbox every weekday morning. Subscribe free at factua.com/blog.
Know someone who should be reading this? Send them to factua.com/blog to subscribe.
