Computer Use
Anthropic's screen-control capability — Claude clicks, types, and scrolls through any GUI app via screenshots and keyboard/mouse primitives.
Capabilities
- screen_control
- mouse_keyboard
- browser_use
Benchmarks
Usage signals
No public usage data recorded yet. See the OpenRouter usage leaderboard for cross-agent comparisons.
Pricing & license
- Pricing
- Anthropic API rates
Base models
Anthropic's previous Opus flagship — superseded by Opus 4.8 on May 28, 2026 (42-day cycle). Optimized for complex reasoning and coding. Improved software engineering, long-running coding tasks, and higher-resolution vision over Claude 4.6.
Anthropic's mid-tier flagship — released 12 days after Opus 4.6 and matching most of its capabilities at ~40% the API cost ($3/$15 per 1M tokens). 1M-token context.
Compare Computer Use with…
Related news
Introducing computer use in Gemini 3.5 Flash
Google introduced computer-use capabilities in Gemini 3.5 Flash, enabling the model to interact with computer interfaces as part of its workflow. It matters because this moves Gemini from text and image generation toward agentic task execution, a step toward automating multi-step actions in software.
Holo3.1: Fast & Local Computer Use Agents
Holo3.1 introduces fast, local computer-use agents designed to operate on-device rather than through cloud-hosted workflows. This matters because local execution can reduce latency, improve privacy, and make interactive agentic automation more practical on consumer hardware.
Codex for (almost) everything
The updated Codex app for macOS and Windows adds computer use, in-app browsing, image generation, memory, and plugins to accelerate developer workflows. These additions push Codex beyond code completion toward a more general agentic tool, with browser and computer control plus persistent memory making it more useful for end-to-end development tasks.
Holotron-12B - High Throughput Computer Use Agent
Holotron-12B is a high-throughput computer-use agent, introducing a 12B-scale model aimed at automating interaction with computers. Its significance is the focus on throughput, which suggests it is designed for fast, scalable agentic workflow execution rather than just isolated task completion.
Introducing GPT-5.4
OpenAI introduced GPT-5.4, describing it as its most capable and efficient frontier model for professional work, with state-of-the-art coding, computer use, tool search, and a 1M-token context. The long context window and improved tool use make it more suited for complex, multi-step workflows, codebases, and document-heavy tasks.