All tasks / Tools
Midscene.js
Open-source vision-based GUI agent and natural-language E2E testing framework for web, mobile and desktop apps.
- web-infra-dev
- Local or cloud
- Free, open source
Where it ranks
- –Browser & computer agentsAgent frameworks · Local · no rated model available here yet
Tools without a benchmark of their own inherit the tier of the best model they offer.
Facts
- Vendor
- web-infra-dev
- Approach
- Agent frameworks
- Where it runs
- Runs fully on your hardware with local models; cloud models optional
- Interface
- API
- Skill level
- Advanced: command line, node graphs or self-hosting
- Price
- Free, open source
- Pricing notes
- Free (MIT); you pay for the model you configure. Supports Qwen3.x, Doubao Seed, DeepSeek, GLM-4.6V, Gemini and self-hostable UI-TARS.
- Source code
- Open source (MIT)
- GitHub stars
- 15.3k (19 Oct 2026)
- Last verified
- 19 Oct 2026
Price from github.com, 19 Oct 2026.
Models it gives you
- –UI-TARS-1.5-7BByteDance Seed · open weights