OpenAI released GPT-6.1 Terra on October 26, a faster mid-sized model built for computer work, and priced it at half of what GPT-5.6 Terra cost. The new GPT-6.1 Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers through the API.
If you pay for ChatGPT, this is the kind of update that quietly changes your day. Your plan stretches further, and the model makes fewer mistakes on everyday tasks.
What GPT-6.1 Terra Is
OpenAI now sells four model classes: Luna, Terra, Sol and Astra, from cheapest to most capable. Terra sits in the middle and aims to balance smarts and cost. It missed last month’s GPT-6 refresh, so this release brings it level with the rest of the family.
OpenAI describes the model as strong at visual reasoning, precise computer use and clearer writing. Think drafting proposals, reading technical drawings, filling out forms and editing video.
GPT-6.1 Terra Pricing and Access
Standard API pricing is $1 per million input tokens and $5 per million output tokens. Cached input costs $0.05 per million tokens. OpenAI says that is 50% below GPT-5.6 Terra and GPT-6.1 Sol.
Inside ChatGPT Work and Codex, the same usage allowance covers more tasks with Terra than with Sol or Astra. Developers call the model as gpt-6.1-terra.
| Detail | GPT-6.1 Terra |
|---|---|
| Input price | $1 per million tokens |
| Cached input | $0.05 per million tokens |
| Output price | $5 per million tokens |
| API name | gpt-6.1-terra |
| Where to use it | ChatGPT Work, Codex, API |
Fewer Errors and Better Coding
OpenAI says it has worked hard this year to cut hallucinations. On its internal test of error-inducing prompts, Terra’s error rate at low reasoning effort fell from 53.6% to 16.7%. At maximum effort it dropped by 18 points to 4.3%.
Coding improved too. On Terminal-Bench 4.0, Terra at maximum effort solves 50.3% of tasks, more than double the 21.5% published for GPT-5.6 Terra. For the toughest coding jobs, OpenAI still points users to Sol and Astra.
- Computer-use scores rise 15 points over GPT-5.6 Terra at about one-third of the cost, per OpenAI.
- Tasks finish over three times faster than on Astra and Sol.
- OpenAI claims about nine times lower cost per task than Claude Opus 5.5 at max effort.
Those comparisons come from OpenAI’s own testing, so treat them as company claims until independent results appear. OpenAI also says Terra nearly matches its top models on a visual reasoning test called MMMU-Pro, at much lower cost.
On OSWorld 2.1, a test of multistep desktop tasks, Terra beats the best GPT-5.6 Terra score while costing far less per task. For anyone running agents all day, that cost gap adds up quickly.
Safety Rating You Should Know
OpenAI says GPT-6.1 Terra meets the Critical threshold in cybersecurity under its Preparedness Framework. It keeps the protections built for GPT-6 Astra.
The company also reports better resistance to prompt injection and a stronger habit of admitting limits instead of bluffing. More detail sits in the system card.
What This Means for You
If you use ChatGPT for daily work, Terra should let you do more before hitting usage limits. Here is how to put it to work.
- Try it first for forms, spreadsheets, invoices and slide drafts.
- Keep Sol or Astra for heavy coding and deep analysis.
- Developers can cut bills on high-volume jobs by switching to the cheaper model.
- Still double-check facts, since errors are lower but not gone.
The release also shows how fast prices are falling for capable AI agents, which makes automation easier for small teams. If you code with AI, our guide to agent tracing shows how teams watch what these tools actually do.
Businesses that read invoices, enter data or review long documents stand to gain the most. OpenAI itself uses invoice entry into accounting software as a prime example of high-volume work that Terra can now automate cheaply.
Sources
Primary source: OpenAI’s GPT-6.1 Terra announcement. Additional reporting from 9to5Mac.
Conclusion
GPT-6.1 Terra gives OpenAI a cheaper, faster middle option that handles real office tasks with fewer mistakes. Expect it to become the default choice for many ChatGPT and API users who do not need top-tier power.
