Small local models
1B–3B parameter models for fast responses, tool-driven workflows, and everyday tasks where latency matters most.
Models
13 local models for iPhone and iPad.
Mid-size models
4B–7B parameter models for richer conversations, stronger reasoning, and longer context without losing the local-first advantage.
OCR from photos and files
OCR extracts text from photos or files on-device, then inserts that text into the model prompt. Model inference stays text-only.
Gemma 4, Llama, Qwen, Mistral, DeepSeek, Phi.
13 models across 6 families. OCR from photos and files is handled via on-device text extraction before prompting. All models are optimized for on-device text inference on iPhone and iPad hardware. See all features →
All models, no paywalls.
Every model in the catalog is available to download and use. No premium model tiers, no per-model unlocks. $2.99 gets you everything. Get SoloLLM →
