OpenAI: computer-use-preview API
computer-use-preview is a specialized preview model by OpenAI designed specifically for computer automation tasks. Unlike general-purpose chat models, it is fine-tuned for the "perception-action" loop of GUI interaction. It can ingest screenshots, interpret visual context, and output precise coordinates for clicks, drags, scrolls, and keystrokes. Primarily accessed via the Responses API, it serves as the reasoning core for AI agents to automate web navigation, form filling, and cross-application workflows.
- Context window: 131,072 tokens
- Max output: 4,096 tokens
- Input: text, image
- Output: text
- Reasoning: Supported
- Tool calling: Supported
- Released: 2025-03-10
- Knowledge cutoff: 2023-09-30
Frequently Asked Questions
What is the context window of computer-use-preview?
computer-use-preview supports a context window of up to 131,072 tokens.
Does computer-use-preview support function calling?
Yes. computer-use-preview supports tool / function calling.
Does computer-use-preview support reasoning?
Yes. computer-use-preview is a reasoning-capable model.
What is the knowledge cutoff of computer-use-preview?
The knowledge cutoff of computer-use-preview is 2023-09-30.