← Back to the blog portfolioEnglish translation of the saved article
English translation of the completed Korean article saved in an earlier personal trial.
The same article is translated and re-typeset for reading. Photos are excluded. No new model generation, factual validation or publishing was performed. This is a translation of past trial output, not current news, current pricing or client delivery.

OpenAI GPT-6.1 Terra,
a work automation model

Key announcement summary

OpenAI has introduced GPT-6.1 Terra, a work-focused model with stronger precise computer control and visual reasoning. What has improved over its predecessor, and what could change in everyday work?

Precise computer use and visual reasoning

GPT-6.1 Terra combines natural writing, precise computer use and detailed visual reasoning in a fast, efficient model. It advances on GPT-5.6 Terra and supports everyday work, from writing proposals and interpreting drawings to filling forms and editing video.

Standard API costs reduced by 50%

Standard API pricing for GPT-6.1 Terra was set 50% below GPT-5.6 Terra and GPT-6.1 Sol. This makes it practical to automate repetitive work at scale, such as reviewing incoming invoices and entering their details into accounting software.


Performance gains and lower error rates by reasoning effort

GPT-5.6 Terra and GPT-6.1 Terra performance comparison

MetricGPT-5.6 TerraGPT-6.1 Terra
Tasks solved at maximum reasoning effort21.5%50.3%
Internal benchmark error rate at low reasoning effort53.6%16.7%

At maximum effort, the error rate falls by 18 percentage points to 4.3%.

At maximum effort, Terra is more than three times as token-efficient as Claude Opus 5.5, while its price per token remains about one quarter as high.

At maximum reasoning effort, the cost per task was reduced by about ninefold.

Platforms and standard API pricing

At a glance

For development teams automating work at scale

GPT-6.1 Terra supports everyday work from writing proposals to editing video while reducing standard API costs. It suits teams building practical automation and reducing the cost of repetitive tasks such as organising invoices and entering data into accounting software.

Sources and original review date

Original record: 2026-10-27 | Primary-source verification marked complete at that time; not revalidated for this translation

Original image notice, translated: the official image came from the OpenAI page and belongs to OpenAI. The article cited it with attribution for introduction and stated that it would be removed on the rights holder’s request. That image is excluded from this reading copy.