AI Models

Smallest possible speech recognition model. Only 32MB. English only. Default speech model - guaranteed to run on any iPhone.

High quality 82M parameter TTS model. Excellent speech synthesis with multiple voice options. 86MB download.

High quality text embedding model. 137M params. Good for RAG and search.

Distilled Whisper. 6x faster than large-v3 with 1% accuracy loss.

Mid-size Whisper model. Strong multilingual speech recognition.

Optimized large Whisper model. Near-best accuracy with faster inference.

Smallest CoreML image generation model. Palettized for minimal size (1.14GB). Runs on any iPhone with 6GB RAM. Default image generation model.

SD 1.5 in single-file GGUF format. Alternative to CoreML. Uses stable-diffusion.cpp with Metal acceleration.

Google's latest tiny 1B model. Excellent quality for its size.

Ultra-compact 1B model. Runs on virtually any device including smartphones.

Largest Whisper model. Best accuracy across all languages and accents.

Ultra-compact vision model. Only 1GB. Answers questions about images.

SD 3 with MMDiT architecture. Superior text rendering.

Meta's compact 3B model designed for edge and mobile deployment.

Higher quality image generation. CoreML optimized for iOS. Requires 6GB+ usable memory (iPad/Mac recommended).

Single-step SDXL. Near-instant image generation.

Tiny but capable 3.8B model. Runs on almost any hardware including phones.

Latest Phi mini with strong reasoning. Drop-in upgrade from Phi-3.5 Mini.

Balanced 4B model with strong reasoning. Great for iPhones.

Vision-language model from Microsoft. Can understand images and documents.

Multimodal vision-language model. Understands images and answers questions about them.

Efficient 7B model from Mistral AI with strong performance for its size.

Efficient 7B model with strong coding and reasoning abilities.

Strong 7B code model rivaling larger coding models. Excellent for local development.

Compact reasoning model. Good reasoning capabilities in a small package.

Meta's 8B parameter instruction-tuned model. Great balance of performance and efficiency for local deployment.

Google's efficient 9B model. Great performance-to-size ratio.

High quality 12B model. Excellent for iPad Pro and Mac.

Mistral's 12B model with excellent instruction following.

Fast 1-4 step generation. State-of-the-art quality. Needs 16GB+ RAM.

Microsoft's 14B parameter model. Punches well above its weight on reasoning.

Strong 14B model with excellent coding and reasoning. iPad Pro recommended.

Powerful 14B code model. Excellent for complex programming tasks.

Google's flagship open model. Near GPT-4 quality. Needs 20GB+ RAM.

Meta's flagship 70B parameter model. Excellent performance rivaling GPT-4 on many benchmarks.

Lightweight TTS voice. High quality English speech synthesis. Default TTS model - runs on any iPhone. Only 63MB.

High quality English male voice. 63MB download. Runs on any device.

Medium quality English voice with natural prosody. 63MB download.

Spanish female voice. Natural prosody.

French female voice.

German male voice.

Chinese Mandarin voice.

Japanese voice.

Korean voice.

Russian female voice.

Portuguese voice.

Italian male voice.

Arabic voice.

Tiny embedding model. Only 23MB. Perfect for on-device search.

Compact English embedding model. Good for basic semantic search.

Compact embedding model from Snowflake. Good multilingual support.

Tiny English reranker. Only 67MB. Use with embedding models for better search.

Tiny multilingual speech recognition. Only 75MB. Supports 99 languages. Runs on any device.

Base whisper model. Good balance of speed and accuracy. 142MB.

English-only base model. Faster and more accurate for English.

Tiny 135M model. Default LLM - guaranteed to run on any iPhone. Only 145MB download. Perfect for quick experiments.

Compact Whisper model. Good accuracy for everyday transcription tasks.

Music generation from text prompts. Requires multiple ONNX files (~435MB total). Experimental iOS support.

High quality English embedding model. Best accuracy for English search.

Compact 360M model. Good for basic tasks on very constrained devices.

Ultra-small 0.5B model from Alibaba. Minimal resource requirements.

Smallest code model. Default code assistant - runs on any iPhone. Great for code completion and simple programming tasks.

Ultra-tiny 500M model. Even smaller than SmolLM. Runs anywhere.

Multilingual reranker. 100+ languages. 1.1GB.

Classic image generation model. Pre-converted to CoreML for iOS/Mac. Downloads as zip, auto-extracts.

SD 2.1 in GGUF format. Better quality than 1.5.

Ultra-compact 1B model from Technology Innovation Institute.

Lightweight 1.1B chat model based on Llama architecture. Great for phones.

Compact code model with strong coding capabilities. Great for mobile coding assistants.

Compact 1.5B model with strong multilingual and coding abilities.

Compact reasoning model distilled from DeepSeek R1. Strong chain-of-thought in a tiny package.

Compact code model with solid code generation and understanding abilities.

Tiny code model. Great for phones. Fast completions.

Capable 1.7B model from HuggingFace. Good balance for mobile devices.

Lightweight code completion model from Google. Fast on-device code suggestions.

Efficient multimodal model with strong image understanding. Optimized for edge devices.

IBM's compact 2B model. Good at following instructions.

Compact vision-language model. Default multimodal model. Can understand images and answer questions about them.

Compact model from LG. Optimized for Korean and English.

Google's compact 2.6B model. Efficient and capable for mobile use.

Versatile 3B model with strong reasoning and multilingual capabilities.

Capable 3B code model. Good balance of coding ability and resource usage.

Compact 3B Falcon model with good performance.

Compact 3B model from Stability AI. Good chat quality for its size.

Fast 3B model tuned for helpful responses.

Code completion model trained on The Stack v2. 600+ languages.

Compact code model with good completion quality.

Google's vision model. Strong at visual QA, captioning, and OCR.

NVIDIA's compact 4B model optimized for edge deployment.

Capable 4B model from H2O.ai. Good for phones.

Efficient 6B bilingual (English/Chinese) model.

Powerful 6.7B code model with excellent code generation across many languages.

Full-size Falcon 3 with strong performance across benchmarks.

Fully open research model. Transparent training.

Fine-tuned Mistral 7B for chat. Strong instruction following.

Larger code model with better completions.

Meta's code-specialized Llama model. Good at code completion.

Strong 7B model from China. Good at tool use and math.

7.8B model from LG. Strong bilingual Korean/English.

IBM's 8B instruction model. Enterprise quality.

Google's instruction-tuned code model. Strong code generation and understanding.

9B bilingual model with strong reasoning.

Strong 9B code model with good reasoning.

10B Falcon model. Good iPad model.

Depth-upscaled 10.7B model. Strong reasoning.

Highest quality FLUX model. 20-50 steps. Mac with 24GB+ RAM.

13B code model for complex tasks. iPad Pro recommended.

22B parameter model. Strong reasoning and multilingual. Needs 16GB+ RAM.

Premium 32B model. Top-tier reasoning. Mac with 32GB+ RAM.