Microsoft's Fara-7B AI agent, rival to GPT‑4o, runs on PC, logs 145k tasks
When Microsoft announced Fara-7B, I was surprised to see a computer-use AI that actually runs on a Windows PC. They pitch it as a “rival” to OpenAI’s GPT-4o, promising everyday software tasks without any cloud connection. That claim immediately makes me wonder how the model was trained and what its real-world performance looks like.
The research team says the system relies on a two-agent setup: one part drafts a plan, the other carries out web-based actions. In their experiments the pair logged about 145,000 successful task trajectories - a number the authors think is enough to teach a smaller model to act on its own. To squeeze that experience into a single model they distilled the data into a 7-billion-parameter version built on Qwen2.5-VL-7B, chosen for its long-context abilities.
The idea is that Fara-7B could cover the same range as GPT-4o while staying entirely on-device.
*In this setup, an "Orchestrator" agent created plans and directed a "WebSurfer" agent to browse the web, generating 145,000 successful task trajectories. The researchers then "distilled" this complex interaction data into Fara-7B, which is built on Qwen2.5-VL-7B, a base model chosen for its long-context capabilities.…
In this setup, an "Orchestrator" agent created plans and directed a "WebSurfer" agent to browse the web, generating 145,000 successful task trajectories. The researchers then "distilled" this complex interaction data into Fara-7B, which is built on Qwen2.5-VL-7B, a base model chosen for its long context window (up to 128,000 tokens) and its strong ability to connect text instructions to visual elements on a screen. While the data generation required a heavy multi-agent system, Fara-7B itself is a single model, showing that a small model can effectively learn advanced behaviors without needing complex scaffolding at runtime.
So, what does Fara-7B actually give you? It’s a 7-billion-parameter Computer Use Agent that you can run straight on a PC, which means you don’t have to lean on huge cloud-hosted models. The developers say it hits top-tier results for its size, with lower latency and better privacy as side benefits.
In their tests an “Orchestrator” agent plotted the steps while a “WebSurfer” agent did the browsing, and together they logged about 145 000 successful task runs. Those runs were then distilled into Fara-7B, which sits on the Qwen2.5-VL-7B base model because that one handles long contexts well. Since the release is still marked experimental, it’s hard to say how quickly companies will take it up, especially when they have to fit it into existing security and compliance stacks.
Still, the design seems to address a big hurdle for businesses: the reliance on massive, cloud-only models. Whether a 7-billion-parameter model can cover the full range of real-world tasks without losing accuracy remains an open question. For now, it looks like a workable, locally-run agent, but we’ll need broader testing before its claims feel solid.
Related Reading
Browse all Microsoft news →
Research & Benchmarks Consensus uses GPT-5 and Responses API to speed scientific research
Research & Benchmarks Developers say Sora, unlike Vine/TikTok, is not about people in social media
Research & Benchmarks Google AI Advisors Let Users Probe Performance with Conversational “Why” Queries
Related Topic Microsoft launches Fara-7B, an agentic Qwen model that solves tasks in ~16 steps
Business & Startups
Related Topic LG's recent webOS update adds Microsoft Copilot app, now removable
AI Tools & Apps
Common Questions Answered
How does Microsoft’s Fara‑7B compare to OpenAI’s GPT‑4o in terms of deployment?
Fara‑7B runs locally on a Windows PC, eliminating the need for cloud servers, whereas GPT‑4o is a cloud‑based service. This on‑device execution gives Fara‑7B lower latency and enhanced privacy compared to GPT‑4o’s remote inference.
What roles do the Orchestrator and WebSurfer agents play in Fara‑7B’s training?
The Orchestrator agent creates detailed action plans, while the WebSurfer agent carries out those plans by browsing the web. Together they generated 145,000 successful task trajectories, which were later distilled into the final Fara‑7B model.
Why was Qwen2.5‑VL‑7B chosen as the base model for Fara‑7B?
Qwen2.5‑VL‑7B provides a 7‑billion‑parameter architecture with a long context window of up to 128,000 tokens and strong text‑to‑visual grounding. These capabilities enable Fara‑7B to understand and manipulate on‑screen elements effectively during computer‑use tasks.
What practical benefits does Fara‑7B claim to offer over larger cloud‑hosted models?
The article notes that Fara‑7B delivers lower inference latency and better privacy because all processing occurs on the user’s PC. Additionally, its 7‑billion‑parameter size achieves state‑of‑the‑art results for a model of this scale, making it a competitive alternative to larger cloud models.