WebBrain: open-source local browser agent with a 450M browser-specialized VLM
ButtercupLyn100 · reddit · 2026-09-29
A developer open-sourced WebBrain, an agent that lets LLMs see and operate a browser, deliberately avoiding dependence on a single cloud provider: it works with local models via LM Studio/Ollama as well as cloud APIs.
The standout is a small vision model webbrain-vl-2-450M, based on LFM-2.5-VL-450M and fine-tuned on browser screenshots/tasks — so browser perception can run locally on a tiny model via WebGPU instead of sending every screenshot to a giant multimodal model.
Architecture: screenshot + browser accessibility tree for perception (not pure DOM parsing), a browser-specialized tiny VLM, a model-agnostic planner, Chrome/Edge/Firefox/Chromium support, optional cloud execution, fully open source. The author wants to explore how far browser agents can go with small local models, and asks which model people would try as the planner.
- Repo: github.com/webbrain-one/webbrain; model and dataset are on Hugging Face.
More from coding & agent
- A naive video transcription fixer pipeline: extract audio+frames, ASR, then correct with a frontier model — capetorch · 2026-09-29
- bdsqlsz is vibe-coding DLSS5 weight training for his in-development 3D game — bdsqlsz · 2026-09-29
- We Built an On-Call Agent That Failed the Right Way — Memory Can Learn the Wrong Lesson — Similar-Split7292 · 2026-09-29
- Extending Jev Mode to Images: Constrained llama.cpp Outputs as Image Selections — opUserZero · 2026-09-29
- Scraping Xiaohongshu hit posts with Codex + a wired Android phone — huangyun_122 · 2026-09-29
- Reverse-Engineering MW2, Minecraft and Skate 3 With Claude and DeepSeek Into One Playable Game — ericcalyborn · 2026-09-29