Artificial Intelligence
OpenAI launches GPT-6.1 Terra
OpenAI launched GPT-6.1 Terra, a model aimed at computer use, visual reasoning and coding, and said it was cheaper and faster than its earlier Terra and Sol models.
- OpenAI launched GPT-6.1 Terra and described it as a model for computer use, visual reasoning, coding and everyday work.
- The company said its API pricing was 50% lower than GPT-5.6 Terra and GPT-6.1 Sol, at $1 per million input tokens and $5 per million output tokens.
- OpenAI said Terra scored 15 percentage points higher than GPT-5.6 Terra on Agents' Last Exam at medium reasoning effort.
- OpenAI said Terra was available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and through the API as gpt-6.1-terra.
OpenAI said it had launched GPT-6.1 Terra, which it positioned as a model for computer use, visual reasoning, coding and routine office tasks. The company said it was intended as a faster and cheaper option for work such as form filling, video editing, reading technical drawings and repetitive desktop operations.
In a blog post, OpenAI said Terra’s API cost was half that of GPT-5.6 Terra and GPT-6.1 Sol. It gave the price as $1 per million input tokens and $5 per million output tokens.
OpenAI said Terra outscored GPT-5.6 Terra by 15 percentage points on Agents' Last Exam at medium reasoning effort. It also said the model was more than three times faster than GPT-6 Astra and GPT-6.1 Sol.
The company said its internal benchmark showed the error rate dropping from 53.6% for GPT-5.6 Terra to 16.7% for GPT-6.1 Terra at low reasoning effort. It said coding results also improved, with Terra scoring 50.3% on Terminal-Bench 4.0 at maximum reasoning effort, while recommending GPT-6.1 Sol or GPT-6 Astra for the hardest coding work.
Named in this story
Companies
- OpenAI
- launched GPT-6.1 Terra and made the performance claims
Products and systems
- GPT-6.1 Terra
- the new model OpenAI announced
- GPT-5.6 Terra
- the earlier model used as a comparison for price and scores
- GPT-6.1 Sol
- a model used in OpenAI’s pricing and speed comparisons
- GPT-6 Astra
- a model used in OpenAI’s pricing and speed comparisons
- Agents' Last Exam
- the benchmark where OpenAI reported a higher score for Terra
- Terminal-Bench 4.0
- the coding benchmark where OpenAI reported Terra’s score
- ChatGPT Work
- one of the places where Terra became available
- Codex
- one of the places where Terra became available
How the source tells it
The register was promotional and press-release like, with OpenAI’s own benchmark claims carrying the story.
- vendor boosterism the piece repeats the company’s self-described performance gains and pricing advantages without independent challenge
- hype the article stacks relative speed, cost and score comparisons to present the model as a major advance