44 AI Models.
All on your device.
From lightning-fast tiny models to powerful vision AI. Choose the perfect model for your needs—all running privately on your iPhone.
Filter by Provider
Filter by Use Case
Filter by Size
⭐ Featured Models
Qwen 3 VL 2B
Analyzes images, reads text from screenshots, and answers questions about photos in 32 languages.
Llama 3.2 3B
Strong reasoning and detailed explanations for complex conversations in 8 languages.
Gemma 3 4B
Top-tier reasoning and 140+ language support with efficient memory usage.
Ministral 3B
Detailed answers and strong instruction following in multiple languages.
Granite 4.0 Micro
Handles complex multi-step tasks with strong instruction following and low memory usage.
SmolLM3 3B
Punches above its weight with strong reasoning and instruction following under 2GB.
LFM2.5 1.2B
Fast and smart in a tiny package. Handles everyday questions and tasks with surprisingly good quality.
Qwen
11 models
Qwen 3 VL 2B
Analyzes images, reads text from screenshots, and answers questions about photos in 32 languages.
Qwen 3 4B
Handles complex questions, multi-step instructions, and conversations in 29+ languages.
Qwen 3 VL 4B
Advanced image analysis for charts, diagrams, documents, and complex visual reasoning tasks.
Qwen 2.5 Coder 3B
Writes, explains, and refactors code in 92+ programming languages.
Qwen 3 4B Thinking
Shows step-by-step reasoning for math, logic puzzles, and complex problems.
Qwen 2.5 3B
Handles translation, creative writing, and longer documents in 29+ languages.
Qwen 3 1.7B
Balanced model for reasoning, logic tasks, and multilingual conversations.
Qwen 2.5 Coder 1.5B
Fast code explanations, bug fixes, and completions in 92+ programming languages.
Qwen 2.5 1.5B
Everyday conversations, writing help, and basic tasks in 29+ languages.
Qwen 3 0.6B
Handles quick questions, simple conversations, and basic tasks with thinking capability.
Qwen 2.5 0.5B
Ultra-fast responses for very simple questions when speed matters most.
Meta
3 models
Llama 3.2 3B
Strong reasoning and detailed explanations for complex conversations in 8 languages.
Llama 3.2 3B Uncensored
Same as Llama 3.2 3B but answers all questions without refusals or content filters.
Llama 3.2 1B
Fast and lightweight for summarization, Q&A, and basic tasks in 8 languages.
Microsoft
2 models
Phi-4 Mini
Excels at math, logic, and reasoning - competes with much larger models on benchmarks.
Phi-3 Mini
Solid math and logic performance trained on high-quality textbook data.
5 models
Gemma 3 4B
Top-tier reasoning and 140+ language support with efficient memory usage.
Gemma 3 1B
Supports 140+ languages in under 1GB - great for less common languages.
Gemma 2 2B
Reliable general chat with precise instruction following.
CodeGemma 2B
Specialized for code completion and fill-in-the-middle tasks.
Gemma 3 270M
Extremely small (253MB) for very basic tasks when storage is critical.
DeepSeek
2 models
DeepSeek R1 1.5B
Shows detailed step-by-step reasoning before answering. Great for learning.
DeepSeek Coder 1.3B
Fast code help in 87 languages. Lightweight but has older training data.
Stability AI
3 models
Stable Code 3B
Writes and explains code in 18 languages with good context for code review.
StableLM Zephyr 3B
Reliable general chat with consistent, helpful responses.
StableLM 2 Zephyr 1.6B
Conversational model supporting 7 European languages under 1GB.
Mistral AI
1 model
Ministral 3B
Detailed answers and strong instruction following in multiple languages.
01.AI
1 model
Yi Coder 1.5B
Fast code completions and debugging in 52 languages under 1GB.
NVIDIA
1 model
Nemotron Mini 4B
Strong chat performance for longer conversations.
IBM
3 models
Granite 4.0 Micro
Handles complex multi-step tasks with strong instruction following and low memory usage.
Granite 4.0 1B
Reliable performance for business and professional tasks under 1GB.
Granite 4.0 350M
Basic tasks with multilingual support at only 223MB.
Hugging Face
4 models
SmolLM3 3B
Punches above its weight with strong reasoning and instruction following under 2GB.
SmolLM2 1.7B
Good conversations and instruction following at just over 1GB.
SmolLM2 360M
Quick responses and basic reasoning at only 390MB.
SmolLM2 135M
Tiny model for quick testing. Download a featured model for better responses.
OpenAI
1 model
GPT-2
Historic GPT-2 from 2019 - the model that started the modern AI wave. Text completion only.
Liquid AI
1 model
LFM2.5 1.2B
Fast and smart in a tiny package. Handles everyday questions and tasks with surprisingly good quality.
Community
6 models
Hermes 3 3B
Roleplay and creative writing with consistent character voices.
Dolphin 2.6 3B
Uncensored model with no content filters for open conversations.
Danube 3 4B
Balanced 4B model with fast responses for general tasks.
Orca Mini 3B
Shows step-by-step reasoning - trained on GPT-4 thinking traces.
TinyLlama 1.1B
Basic chat under 700MB - runs fast on almost any device.
Danube 3 500M
Ultra-fast responses for basic tasks with multilingual support.
Ready to try these models?
Download KernelAI and start chatting with any of these 44 models privately on your device.
Download KernelAI