ByteDance Open-Sources 7B Desktop GUI Control Agent
ByteDance open-sources a 7B desktop GUI control model capable of operating any application without app-specific training — 10k GitHub stars on day one.
ByteDance open-sources a 7B desktop GUI control model capable of operating any application without app-specific training — 10k GitHub stars on day one.
Microsoft's Phi-Ground-Any is a 4B vision model achieving SOTA on GUI grounding benchmarks — enabling AI agents to precisely click screen elements.
Microsoft Research's 1,000 Synthetic Computers provides long-horizon agent training environments scaling to billions of virtual worlds without user telemetry.
OpenAI's Codex for Work brings role-based AI assistance to docs, slides, and spreadsheets—Computer Use now runs 42% faster than before.
OpenAI's Chronicle research preview (April 20) captures your screen periodically to build ambient memory that makes Codex smarter over time — not available in EU/UK/CH.

GPT-5.5 scores 2.5× better intelligence-per-token than 5.4, surpasses the human baseline on OS World, and expands Codex into a full desktop agent.
OpenAI upgrades Codex with parallel background computer-use agents, an in-app click-to-comment browser, native image gen, and a 90+ plugin marketplace.
Curated AI insights — sent when there's something worth your inbox.