{"models":[{"id":"bonsai-27b","family":"prismml","task":"vision-language","parameters":27300000000,"description":"PrismML Bonsai 27B 1-bit vision-language model for local reasoning.","variants":["macos-arm64"]},{"id":"functiongemma-270m-it","family":"functiongemma","task":"text-generation","parameters":270000000,"description":"Google FunctionGemma 270M instruction-tuned function-calling model for local inference.","variants":["universal"]},{"id":"gemma-3-1b","family":"google","task":"text-generation","parameters":1000000000,"description":"Gemma 3 1B - Google's mobile-optimized instruction-tuned LLM with 32K context","variants":["universal"]},{"id":"gemma-4-e2b","family":"google","task":"text-generation","parameters":5100000000,"description":"Gemma 4 E2B - Google's compact multimodal LLM with 2.3B effective params, 128K context, and audio/image/video understanding","variants":["universal"]},{"id":"gemma-4-e4b","family":"google","task":"text-generation","parameters":8000000000,"description":"Gemma 4 E4B - Google's mid-range multimodal LLM with 4.5B effective params, 128K context, and audio/image/video understanding","variants":["universal"]},{"id":"gemma3npc-1b","family":"google","task":"text-generation","parameters":1000000000,"description":"Gemma3NPC 1B - Mobile-friendly NPC roleplay fine-tune of Gemma 3 1B for in-character game dialogue","variants":["universal"]},{"id":"gemma3npc-it","family":"google","task":"text-generation","parameters":7000000000,"description":"Gemma3NPC-it - NPC roleplay fine-tune of Gemma3n-E4B for in-character game dialogue","variants":["universal"]},{"id":"kitten-tts-micro-0.8","family":"kittenml","task":"text-to-speech","parameters":40000000,"description":"KittenTTS Micro 0.8 - Compact StyleTTS 2 (40M params, 8 named voices, OpenPhonemizer)","variants":["universal"]},{"id":"kitten-tts-mini-0.8","family":"kittenml","task":"text-to-speech","parameters":80000000,"description":"KittenTTS Mini 0.8 - High-quality StyleTTS 2 (80M params, 8 named voices, OpenPhonemizer)","variants":["universal"]},{"id":"kitten-tts-nano-0.2","family":"kittenml","task":"text-to-speech","parameters":15000000,"description":"KittenTTS Nano 0.2 - Ultra-lightweight TTS (15M params, <25MB)","variants":["universal"]},{"id":"kitten-tts-nano-0.8","family":"kittenml","task":"text-to-speech","parameters":15000000,"description":"KittenTTS Nano 0.8 - Ultra-lightweight StyleTTS 2 (15M params, 8 named voices, OpenPhonemizer)","variants":["universal"]},{"id":"kokoro-82m","family":"hexgrad","task":"text-to-speech","parameters":82000000,"description":"Kokoro 82M - High-quality TTS with 24 voices, Misaki dictionary","variants":["universal"]},{"id":"lfm2-vl-450m","family":"liquid","task":"vision-language","parameters":450000000,"description":"Liquid LFM2-VL 450M compact vision-language model for edge inference.","variants":["universal"]},{"id":"lfm2.5-1.2b-instruct","family":"liquid","task":"text-generation","parameters":1170000000,"description":"Liquid LFM2.5 1.2B Instruct - hybrid conv+attention LLM optimized for edge deployment (agentic tasks, data extraction, RAG, tool calling)","variants":["universal"]},{"id":"lfm2.5-1.2b-thinking","family":"liquid","task":"text-generation","parameters":1170000000,"description":"Liquid LFM2.5 1.2B Thinking - hybrid conv+attention reasoning LLM optimized for edge deployment (fast CPU inference, low memory)","variants":["universal"]},{"id":"lfm2.5-230m","family":"liquid","task":"text-generation","parameters":230000000,"description":"Liquid LFM2.5 230M - smallest hybrid conv+attention LLM optimized for edge deployment (9 languages, tool calling, fast CPU inference)","variants":["universal"]},{"id":"lfm2.5-350m","family":"liquid","task":"text-generation","parameters":354000000,"description":"Liquid LFM2.5 350M - hybrid conv+attention LLM optimized for edge deployment (9 languages, tool calling)","variants":["universal"]},{"id":"lfm2.5-vl-3b","family":"liquid","task":"vision-language","parameters":3000000000,"description":"Liquid LFM2.5-VL 3B vision-language model for local inference.","variants":["universal"]},{"id":"llama-3.2-1b","family":"meta","task":"text-generation","parameters":1000000000,"description":"Llama 3.2 1B - Meta's lightweight mobile-optimized instruction-tuned LLM with 128K context","variants":["universal"]},{"id":"ministral-3-3b","family":"mistral","task":"text-generation","parameters":3400000000,"description":"Ministral 3 3B Instruct - Mistral AI's edge-optimized instruction-tuned LLM with 256K context","variants":["universal"]},{"id":"mistral-7b","family":"mistral","task":"text-generation","parameters":7000000000,"description":"Mistral 7B Instruct v0.3 - High-quality desktop LLM with function calling support","variants":["universal"]},{"id":"neutts-air-q4","family":"neuphonic","task":"text-to-speech","parameters":500000000,"description":"NeuTTS Air (~500M) — codec TTS with voice cloning. Q4 GGUF backbone + NeuCodec decoder. Higher quality than Nano.","variants":["universal"]},{"id":"neutts-nano-q4","family":"neuphonic","task":"text-to-speech","parameters":120000000,"description":"NeuTTS Nano (120M) — codec TTS with voice cloning. Q4 GGUF backbone + NeuCodec ONNX decoder.","variants":["universal"]},{"id":"phi4-mini","family":"microsoft","task":"text-generation","parameters":3800000000,"description":"Phi-4 Mini 3.8B - Microsoft's compact reasoning LLM with 128K context","variants":["universal"]},{"id":"phonon-31m-en","family":"gradium","task":"text-to-speech","parameters":31000000,"description":"Phonon 31M - English streaming TTS from Gradium (24 kHz, 38 built-in voices), one GGUF for the zzz CPU engine","variants":["android-arm64","linux-x86_64","macos-arm64"]},{"id":"qwen2.5-0.5b-instruct","family":"qwen","task":"text-generation","parameters":500000000,"description":"Qwen 2.5 0.5B Instruct - Small but capable instruction-tuned LLM from Alibaba Cloud","variants":["universal"]},{"id":"qwen3.5-0.8b","family":"qwen","task":"text-generation","parameters":800000000,"description":"Qwen 3.5 0.8B - Alibaba Cloud's lightweight multimodal LLM (text-only mode, 201 languages)","variants":["universal"]},{"id":"qwen3.5-2b","family":"qwen","task":"text-generation","parameters":2000000000,"description":"Qwen 3.5 2B - Alibaba Cloud's compact multimodal LLM (text-only mode, 201 languages)","variants":["universal"]},{"id":"smollm2-360m","family":"huggingface","task":"text-generation","parameters":360000000,"description":"SmolLM2 360M - HuggingFace's best tiny LLM, excellent quality/size ratio","variants":["universal","web"]},{"id":"wav2vec2-base-960h","family":"facebook","task":"speech-recognition","parameters":95000000,"description":"Wav2Vec2 Base 960h - English ASR with CTC decoding","variants":["universal"]},{"id":"whisper-base-en-ggml","family":"openai","task":"speech-recognition","parameters":74000000,"description":"Whisper Base English - English-only ASR (whisper.cpp/GGML runtime)","variants":["ios-arm64","universal"]},{"id":"whisper-base-ggml","family":"openai","task":"speech-recognition","parameters":74000000,"description":"Whisper Base - Multilingual ASR (whisper.cpp/GGML runtime)","variants":["ios-arm64","universal"]},{"id":"whisper-large-v3-turbo-ggml","family":"openai","task":"speech-recognition","parameters":809000000,"description":"Whisper Large v3 Turbo - Highest-accuracy multilingual ASR (whisper.cpp/GGML runtime)","variants":["universal"]},{"id":"whisper-medium-ggml","family":"openai","task":"speech-recognition","parameters":769000000,"description":"Whisper Medium - Multilingual ASR (whisper.cpp/GGML runtime)","variants":["universal"]},{"id":"whisper-small-ggml","family":"openai","task":"speech-recognition","parameters":244000000,"description":"Whisper Small - Multilingual ASR (whisper.cpp/GGML runtime)","variants":["universal"]},{"id":"whisper-tiny","family":"openai","task":"speech-recognition","parameters":39000000,"description":"Whisper Tiny - Fast multilingual ASR (Candle/SafeTensors runtime)","variants":["universal"]},{"id":"whisper-tiny-en-ggml","family":"openai","task":"speech-recognition","parameters":39000000,"description":"Whisper Tiny English - Fastest English-only ASR (whisper.cpp/GGML runtime)","variants":["ios-arm64","universal"]},{"id":"whisper-tiny-ggml","family":"openai","task":"speech-recognition","parameters":39000000,"description":"Whisper Tiny - Fast multilingual ASR (whisper.cpp/GGML runtime)","variants":["ios-arm64","universal"]}]}