Skip to content
ybrid
Models
Leaderboard
Demo
Blog
Docs
Console
526
Sign in
Menu
Models
Leaderboard
Demo
Blog
Docs
Console
GitHub
526
On-device AI
for any scale.
Fast local inference towards unlimited tokens
GitHub
Copy prompt
Paste to agent
Copy failed
The Xybrid logo becomes a throughput comparison: 33.25 tokens per second for xybrid engine and 19.62 for llama.cpp, then returns to the logo.