News - GPT-6.1 Terra — OpenAI Makes Computer-Using AI Agents Faster and Cheaper
OpenAI has released GPT-6.1 Terra, a new model optimized for computer use, visual reasoning and professional work, with a strong emphasis on efficiency and lower execution costs.
Unlike frontier models primarily optimized for complex reasoning, Terra targets practical workflows such as interpreting documents, navigating application interfaces, filling forms and completing multistep tasks across software applications.
Key technical facts
- Model:
gpt-6.1-terra - Context window: 1,050,000 tokens
- Maximum output: 128,000 tokens
- Capabilities: Visual reasoning, computer use, coding and writing
- API pricing: $1 per million input tokens / $5 per million output tokens
- Cached input: $0.05 per million tokens
- Availability: OpenAI API, ChatGPT Work and Codex
OpenAI reports that GPT-6.1 Terra achieves significantly better results than GPT-5.6 Terra on computer-use benchmarks, while offering 50% lower API pricing.
On Agents' Last Exam, OpenAI reports that Terra can approach GPT-6 Astra's performance at roughly one-tenth of its token price, depending on the configuration.
Why does it matter?
AI agents increasingly execute workflows involving hundreds of model calls, tool interactions and application operations.
For these workloads, cost per successfully completed task can matter more than raw model intelligence.
Terra represents an important architectural direction: use efficient models for routine computer interaction and reserve expensive frontier reasoning models for genuinely difficult problems.
Summary: The future of practical AI agents may depend less on using the smartest model everywhere and more on using the right model for each task.