One dollar per million input tokens is now the price of an OpenAI model that can read a screen, click buttons and type into business software. OpenAI released GPT-6.1 Terra on October 26, 2026, as a faster and cheaper model for everyday computer work, visual reasoning and writing. It replaces GPT-5.6 Terra, costs $1 per million input tokens and $5 per million output tokens in the API, and is available in ChatGPT Work and Codex for paying plans.
Three days later, on October 29, OpenAI added monthly API credits to ChatGPT Pro and Business plans. Together, the two moves cut the cost of trying AI agents on real office tasks, which is why this launch matters more to finance teams, agencies and developers than most model updates.
What Professionals Get From GPT-6.1 Terra
- API pricing is $1 per million input tokens, $0.05 for cached input and $5 for output, which OpenAI says is half the standard price of GPT-5.6 Terra.
- Terra sits below GPT-6.1 Sol and GPT-6 Astra and above Luna in OpenAI’s lineup.
- In ChatGPT it appears in Work and Codex for Plus, Pro, Business, Enterprise and Edu users, not in regular chat.
- OpenAI rates Terra “Critical” for cybersecurity capability and gives it the same safeguards as GPT-6 Astra.
- From October 29, Pro and Business subscribers can claim monthly API credits worth up to $500.
Where GPT-6.1 Terra Fits in OpenAI’s Model Lineup
OpenAI now sells several model sizes under space-themed names. Luna is the smallest. Terra is the mid-sized workhorse. Sol is the flagship for heavier work, and GPT-6 Astra is the most capable model in the range. OpenAI still recommends Sol or Astra for in-depth research and the hardest coding jobs. Terra is pitched at the large middle band of tasks where speed and price matter more than peak reasoning.
The API model ID is gpt-6.1-terra. It takes text and images as input and returns text, and it supports function calling, structured outputs and OpenAI’s hosted computer-use tool.
Office Tasks OpenAI Built Terra to Handle
Computer use means the model looks at a screenshot, works out what is on the screen, and then clicks, types or switches apps the way a person would. That matters for software with no API, such as an old accounting package or a supplier portal. OpenAI’s launch examples include filling in forms, logging invoices into accounting software, drafting proposals, reading technical drawings and editing video. In one demo the model turned architectural drawings into a 3D model.
For a small business, the invoice example is the practical one. A bookkeeper could ask Terra to open each emailed bill, read the amounts and enter them into the ledger, then check the work. Firms already linking AI to their books, such as through the Xero and Gemini integration, will recognise the pattern. Terra’s appeal is that it can do this through the screen when no integration exists.
GPT-6.1 Terra API Pricing Against Sol
Agent work burns tokens because every step sends screenshots and context back to the model. Price per token therefore decides whether an automation pays for itself. The figures below are OpenAI’s standard API prices at launch.
| Model | Input per 1M tokens | Cached input per 1M tokens | Output per 1M tokens |
|---|---|---|---|
| GPT-6.1 Terra | $1.00 | $0.05 | $5.00 |
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 |
We priced a sample month for one automation that sends 10 million input tokens and receives 1 million output tokens. With no caching, Terra costs $15 and Sol costs $30. If 8 million of those input tokens are repeated instructions served from cache, Terra falls to $7.40 and Sol to $14.80.
| Scenario | GPT-6.1 Terra | GPT-6.1 Sol |
|---|---|---|
| 10M input, 1M output, no cache | $15.00 | $30.00 |
| 2M fresh input, 8M cached, 1M output | $7.40 | $14.80 |
The lesson is that caching matters almost as much as model choice. Keep long system prompts and reference files identical between calls so they qualify as cached input. If you bill clients for automation work, a tool such as our freelance hourly rate calculator helps you set a price that covers API costs as well as your time.
OpenAI’s Benchmark Claims for Terra, Read Carefully
All of the performance figures so far come from OpenAI, and none has been checked independently yet. OpenAI says Terra:
- Beats its predecessor: it scored 15 percentage points higher than GPT-5.6 Terra on one test at about a third of the cost.
- Matches the top model on some work: it equalled GPT-6 Astra on one evaluation at about a tenth of the price.
- Makes fewer factual errors: its error rate on trick questions at the lowest setting fell from 53.6% to 16.7%.
- Codes better in the terminal: it solved about half the tasks in a command-line coding test, more than double the older model.
OpenAI also says Terra finishes tasks more than three times faster on average than Astra and Sol, and that it completed tasks at roughly one-ninth the cost of Anthropic’s Claude Opus 5.5. Treat these as vendor claims. A sensible test is to run Terra on 20 of your own routine tasks and compare its error rate with what your team accepts today.
Who Can Use Terra in ChatGPT Work and Codex
Terra rolled out on October 26 in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu plans. It is picked from the model list and is not offered in the regular chat view. Free and Go users do not get it at launch, and Enterprise or Edu admins may need to switch it on. Developers can call it through the API, and Codex releases 0.165.0 and 0.166.0 added it to the model catalogues.
API Credits Now Bundled With Pro and Business Plans
The second piece of news arrived on October 29. Eligible subscribers can now claim promotional API credits each month, separate from their ChatGPT usage. OpenAI also doubled the Work and Codex allowance on its Pro 100 and Pro 500 plans and dropped a planned cut to Pro 200.
| Plan | Monthly API credits |
|---|---|
| ChatGPT Pro | $100, $200 or $500, depending on the Pro tier |
| ChatGPT Business, Standard seat | $20 per seat |
| ChatGPT Business, Premium seat | $100 per seat |
| Business workspace cap | Pooled up to $500 |
At Terra’s prices, $100 in credits covers our no-cache sample month more than six times over. For a small agency, that is enough to prototype an agent before anyone signs off a separate API budget.
Terra’s Cyber Rating and the Guardrails Around Computer Use
Under OpenAI’s Preparedness Framework, Terra reaches the “Critical” level for cybersecurity skills, so it runs with the same protections as GPT-6 Astra. OpenAI says the model is more likely than its predecessor to stay within its assigned task and to admit what it cannot do. For businesses, an agent that can click through systems should get its own login with limited rights, never an admin account. Our small business cybersecurity checklist and our guide to writing an AI acceptable use policy cover those controls.
Start Terra on Dull, Checkable Office Work
We see Terra as the first OpenAI computer-use model priced for routine back-office work. The halved price, the bundled credits and its place inside ChatGPT Work mean a non-developer can try it on real tasks this week. The sensible starting points are dull, repetitive jobs with an easy check at the end, such as invoice entry, form filling and status reports. We would keep a person approving anything that sends money, signs a contract or emails a client, at least until independent tests confirm OpenAI’s accuracy figures.
Quick Answers Before Your Team Tries Terra
When Was GPT-6.1 Terra Released?
OpenAI released GPT-6.1 Terra on October 26, 2026, in the API and in ChatGPT Work and Codex.
How Much Does GPT-6.1 Terra Cost in the API?
$1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens, half the standard price of GPT-5.6 Terra.
Is GPT-6.1 Terra Available on the Free ChatGPT Plan?
No. At launch it is offered in ChatGPT Work and Codex to Plus, Pro, Business, Enterprise and Edu users only.
What Is the Difference Between GPT-6.1 Terra and GPT-6.1 Sol?
Sol is the larger flagship model and costs twice as much per token. OpenAI recommends Sol or GPT-6 Astra for complex coding and in-depth research, and Terra for fast, high-volume computer work and writing.
