$ less news/make-money-with-chatgpt-on-gpt-6-1-terra-check-your-preset.md

Make money with ChatGPT on GPT-6.1 Terra: cheaper per task, so check what your preset switched to

2026-10-27

If you make money with ChatGPT through an agent, the model under it may have changed overnight. On October 26 OpenAI started rolling out GPT-6.1 Terra in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu plans. OpenAI says your plan's overall usage limits are unchanged, but the same allowance covers more tasks with Terra than with GPT-5.6 Terra, and substantially more than with GPT-6.1 Sol or GPT-6 Astra.

The line in the release notes I'd flag in a bug report: some Plus and Pro users may now see GPT-6.1 Terra at Medium instead of GPT-6.1 Sol at Light in the default Power presets in Work. If you never picked a model by hand, your agent may already be running a different one than it was last week. On the API side, OpenAI lists Terra at 50% below GPT-5.6 Terra and GPT-6.1 Sol, which BleepingComputer reports as $1 per million input tokens and $5 per million output tokens.

OpenAI also published its own factuality numbers. On an internal set of error-inducing prompts, Terra's error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7%, and to 4.3% at maximum effort. Better, but at low effort that's still roughly one wrong answer in six on prompts built to trip it, and paid research tasks get checked. OpenAI also says Sol and Astra remain the better choice for the most complex coding work.

Changelog note on my setup: nothing to edit. The earning prompt is still one line and the rules live on /setup, so a model swap doesn't touch it. What changes is the cost side of each task. My test plan after a swap like this: 1) open the model list and write down which model and effort your agent is actually on; 2) run one paid task on Terra at Medium and one on whatever you used before; 3) compare usage burned against sats credited for each; 4) read both reports for wrong facts before you trust the cheaper one; 5) don't drop to low effort to save usage on paid work.

If the cheaper run wins, the sats still need somewhere safe to land. Marcus wrote up why he'd keep a receive-only key on the wallet your agent earns into, and it pairs well with a model that's doing more of the clicking for you.

Why ubiprompt is the best place for money-making prompts on a launch day like this: I treat the prompt like code. It's one versioned line, the behaviour rules are written down on a public setup page, and when the model underneath changes I publish a test plan instead of a hype post. I don't publish earnings I can't show, mine or anyone's.

— priya · get the prompt

If you're wiring this up for the first time: pointing your AI agent at neetbros.