GPT-6.1 Terra: OpenAI Cuts AI Automation Prices in Half | Freewares

The AI price war just got another escalation. OpenAI has launched GPT-6.1 Terra, a model built for automating work, at $1 per million input tokens and $5 per million output tokens. That is half the price of GPT-5.6 Terra, which held the same role until now.

The model is available in the OpenAI API, ChatGPT Work, and Codex, with deployment also landing on GitHub Copilot, Amazon Bedrock, and Microsoft Foundry. Context window: 1,050,000 tokens, enough to swallow a small codebase or a year of meeting notes in one go.

Cheaper intelligence changes the math

A 50% price cut sounds like a footnote until you multiply it by the workloads companies are actually running. Automation that was too expensive to bother with last quarter starts to pencil out this one. That is the real story here: not a smarter model, but a cheaper one, aimed squarely at the routine work businesses pay people or software to do today.

This could put pressure on competing providers to lower their prices too.

Where free and local still wins

Before you rent your workflow by the million tokens, the usual reminder applies. For a growing list of everyday jobs, summarizing notes, drafting text, cleaning up data, free, open-weight models can run on your own machine, keep your data local as long as your setup does not send it to external services, and are getting good fast. Tools like LM Studio or Ollama make that a download, not a project.

The honest split: local models win on privacy, predictable cost, and anything you would rather not send to a server. Paid APIs still win on raw capability and zero setup. GPT-6.1 Terra just moved the price line, not the logic.