Kanso Koi and Satori models for dictation formatting and text refinement on Apple Silicon.
AI & ML interests
20 Year Old Student • Indie Developer AI • Code • Ambitious Ideas
Recent Activity
View all activity
Organization Card
Hi, we're BuilderHelm 👋
We're an independent AI studio building small, personalized models and practical tools for everyday work. Our focus is private, on-device AI, efficient inference, and tools that keep people in control of their agents.
We build in public and share our work through code, model releases, and reproducible benchmarks. When we build on an open-source model, our derivative stays open source, with credit to the original creators.
Our models
- Speech: Parakeet CTC 110M GGUF · Whisper Hindi2Hinglish Prime GGML
- Kanso Koi: 1.0 · 0.9 · 0.8
- Kanso Satori: 1.0 · 1.0 BF16 · 1.0 3-bit · 0.9
Explore ZenVoice, our private, on-device dictation app for Mac, and follow our work on GitHub and Twitter / X.
models 12
BuilderHelm/Parakeet-TDT-0.6B-V3-CoreML
Automatic Speech Recognition • Updated • 111
BuilderHelm/Parakeet-TDT-CTC-110M-CoreML
Automatic Speech Recognition • Updated • 276
BuilderHelm/Parakeet-CTC-110M-CoreML
Automatic Speech Recognition • Updated • 191
BuilderHelm/Kanso-Koi-1.0
Text Generation • 2B • Updated • 233
BuilderHelm/Kanso-Satori-1.0-3bit
Text Generation • 4B • Updated • 227
BuilderHelm/Kanso-Satori-1.0
Text Generation • 4B • Updated • 222
BuilderHelm/Kanso-Satori-1.0-bf16
Text Generation • 4B • Updated • 397
BuilderHelm/Kanso-Koi-0.8
Text Generation • 2B • Updated • 696 • 1
BuilderHelm/Kanso-Koi-0.9
Text Generation • 2B • Updated • 229
BuilderHelm/Whisper-Hindi2Hinglish-Prime-GGML
Automatic Speech Recognition • Updated • 2