GPT-6.1 Terra (Medium)

GPT-6.1 Terra is a mid-tier multimodal language model developed by OpenAI, engineered specifically to optimize the balance between speed, cost-efficiency, and advanced performance across high-volume professional workflows. Positioned as a significant step up from its predecessor, GPT-5.6 Terra, the model combines rapid processing speeds with enhanced factuality, natural writing, and cost-effective operations.

Key Capabilities and Architecture

The model is built to handle robust data inputs, accepting both text and images to drive automated workflows. Its standout feature is precise computer use, allowing it to parse application screens, interpret complex technical drawings, and navigate digital interfaces. Furthermore, it supports customizable reasoning effort levels ranging from none to max, letting developers fine-tune latency and execution costs depending on the complexity of the task.

Context Window and Performance

GPT-6.1 Terra features a massive 1,050,000-token context window with a maximum input capacity of 922,000 tokens and an output limit of 128,000 tokens. This expansive context allows it to process large documents, lengthy codebases, and extensive visual datasets efficiently. Designed for cost-sensitive environments, it delivers substantial token efficiency and competitive benchmark results, particularly in visual reasoning and agentic desktop execution tasks.