OpenAI launches GPT-6.1 Terra, its frontier model at half the price
OpenAI announced GPT-6.1 Terra on October 26, 2026, the newest model in its Terra line, and the practical news is the price: standard API rates are 50% lower than both GPT-5.6 Terra and GPT-6.1 Sol. The company frames it as its cost-intelligence frontier, meaning frontier-class capability aimed at high-volume work rather than the hardest possible tasks.
The price is the story
Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens. That is half of what OpenAI charges for GPT-6.1 Sol. Its own example of the kind of work this unlocks is the sort that is tedious at scale: reading incoming invoices and entering their details into accounting software.
The discount is not only for API customers. In ChatGPT Work and Codex, the same usage allowance covers more tasks with GPT-6.1 Terra than with GPT-5.6 Terra, and, per the announcement, substantially more than with GPT-6.1 Sol or GPT-6 Astra.
From visual understanding to action
The second part of the pitch is agentic computer use. OpenAI reports GPT-6.1 Terra clearing the Agents' Last Exam more than three times faster than its predecessor, and it publishes results on OSWorld 2.1 and MMMU-Pro, benchmarks that test whether a model can ground what it sees on a screen into a sequence of actions. For teams running agents over desktop or browser environments, that is the capability to watch.
Factuality and coding
OpenAI says it spent the year cutting hallucinations and factual errors. On its internal benchmark of error-inducing prompts, the error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra, a drop of nearly 37 percentage points. At maximum effort the rate fell by 18 points to 4.3%, roughly matching GPT-6 Astra and GPT-6.1 Sol.
On coding, the company describes the step up over GPT-5.6 Terra in software engineering as substantial, citing Terminal-Bench 4.0 and Terminal-Bench-Science 0.1. All of these numbers come from OpenAI's own evaluations, so treat them as vendor-reported until an independent replication lands.
Safety and availability
GPT-6.1 Terra meets the Critical threshold in cybersecurity under OpenAI's Preparedness Framework and keeps the protections developed for GPT-6 Astra. The company also reports the GPT-6 series alignment gains carry over: Terra is more likely to stay in scope and to honestly disclose its limitations, and it showed strong resistance to prompt-injection attempts in its tests. A system card accompanies the release with the details.
From launch, the model is available to all Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and through the API as gpt-6.1-terra. For a team whose agent bill is dominated by volume, the obvious move is to run Terra against Sol on its own workloads: same family, same safety tier, half the sticker.
Source: openai.com
More news
Have something in mind?
Start a conversation