# ZETIC Melange Documentation

> ZETIC Melange is an on-device AI SDK that automates NPU acceleration for iOS and Android.
> Deploy any AI model with 3 lines of code — automatically optimized for CPU, GPU, and NPU.

## Docs

- [ZETIC Melange](/): Deploy any AI model to iOS and Android, automatically optimized for CPU, GPU, and NPU.
- [Cache Management](/api-reference/cache-management): TBD reference for managed model cache behavior and ModelCacheManager.
- [Error Codes](/api-reference/error-codes): Reference for ZETIC Melange error codes and their meanings.
- [Custom Preprocessing](/how-to-guides/custom-preprocessing): Implement custom input preprocessing for your Melange models.
- [Inference Mode Selection](/how-to-guides/inference-mode-selection): Choose between CPU, GPU, and NPU inference modes in ZETIC Melange.
- [Multi-Model Pipelines](/how-to-guides/multi-model-pipeline): Chain multiple AI models together on-device with ZETIC Melange.
- [Performance Optimization](/how-to-guides/performance-optimization): Tips for optimizing on-device AI performance with ZETIC Melange.
- [Key Concepts](/introduction/key-concepts): Essential terminology and concepts for working with ZETIC Melange.
- [Platform Support Matrix](/introduction/platform-support-matrix): Supported platforms, devices, chipsets, and model formats for ZETIC Melange.
- [Pricing and Plans](/introduction/pricing-and-plans): ZETIC Melange pricing tiers: Free, Pro, Pro+, and Enterprise.
- [What is Melange?](/introduction/what-is-melange): Learn what ZETIC Melange is, who it's for, and how it automates on-device AI deployment.
- [Function Calling](/llm-inference/function-calling): Connect ZeticMLange LLM output to app-defined tools on Android, iOS, and Flutter.
- [LLM Inference Modes](/llm-inference/inference-modes): Configure LLM inference modes for speed or accuracy with ZETIC Melange.
- [LLM Inference Overview](/llm-inference/overview): Run large language models on-device with ZETIC Melange.
- [Retrieval-Augmented Generation](/llm-inference/rag): Use retrieval and local context with ZeticMLange LLMs.
- [Streaming Token Generation](/llm-inference/streaming-generation): Stream LLM tokens in real-time on Android, iOS, and Flutter with ZETIC Melange.
- [Supported LLM Models](/llm-inference/supported-models): List of supported large language models for on-device inference with Melange.
- [Vision-Language Inference](/llm-inference/vision-language): Run image and text prompts with ZeticMLange LFM-VL models.
- [Web Dashboard](/model-deployment/dashboard): Deploy AI models through the ZETIC Melange web dashboard.
- [Performance-Adaptive Deployment](/model-deployment/performance-adaptive-deployment): How Melange automatically selects the optimal inference path for each device.
- [Understanding Model Keys](/model-deployment/understanding-model-keys): How model keys and personal keys work in ZETIC Melange.
- [Pre-built Models](/model-preparation/hugging-face-models): Use ready-to-run models from the Melange Dashboard and Hugging Face without uploading your own.
- [ONNX Models](/model-preparation/onnx): Convert TensorFlow, Keras, and scikit-learn models to ONNX for ZETIC Melange.
- [PyTorch Export](/model-preparation/pytorch-export): Export PyTorch models for ZETIC Melange using Exported Program.
- [Supported Model Formats](/model-preparation/supported-formats): Model formats supported by ZETIC Melange: PyTorch and ONNX.
- [Benchmark Methodology](/performance/benchmark-methodology): How ZETIC Melange benchmarks are measured across 200+ physical devices.
- [Benchmarks](/performance/benchmarks): Real-world performance benchmarks for ZETIC Melange across 200+ devices.
- [Device Support](/performance/device-compatibility): ZETIC Melange device support and NPU acceleration.
- [Quick Start: Your First Inference in 5 Minutes](/quick-start): Get your first on-device AI inference running in 5 minutes with ZETIC Melange.
- [Prerequisites](/quick-start/prerequisites): What you need before getting started with ZETIC Melange.
- [Upload Your Model](/quick-start/upload-your-model): Upload and deploy your own AI model with ZETIC Melange.
- [Changelog](/release-notes/changelog): ZETIC Melange SDK release notes and version history.
- [Migration Guide: MLange → Melange](/release-notes/migration-mlange-to-melange): What changed in the rebrand from MLange to Melange, and what you need to update.
- [Community](/resources/community): Join the ZETIC Melange developer community.
- [GitHub Repositories](/resources/github): ZETIC Melange open-source repositories and sample apps.
- [Support](/resources/support): Get help with ZETIC Melange.
- [Android Issues](/troubleshooting/android-issues): Troubleshoot common Android integration issues with ZETIC Melange.
- [Common Errors](/troubleshooting/common-errors): Solutions to the most common ZETIC Melange integration errors.
- [FAQ](/troubleshooting/faq): Frequently asked questions about ZETIC Melange.
- [iOS Issues](/troubleshooting/ios-issues): Troubleshoot common iOS integration issues with ZETIC Melange.
- [Model Conversion Issues](/troubleshooting/model-conversion-issues): Troubleshoot model conversion and upload issues with ZETIC Melange.
- [Audio Classification (YAMNet)](/tutorials/audio-classification-yamnet): Classify 521 audio event types on-device using YAMNet with ZETIC Melange.
- [Face Detection](/tutorials/face-detection): Run MediaPipe Face Detection on-device with ZETIC Melange.
- [Face Emotion Recognition](/tutorials/face-emotion-recognition): Classify facial emotions on-device using EMO-AffectNet with ZETIC Melange.
- [Face Landmark Detection](/tutorials/face-landmark): Detect facial landmarks using a two-model pipeline with ZETIC Melange.
- [Object Detection (YOLOv8 / YOLOv11)](/tutorials/object-detection-yolo): Build on-device object detection with YOLOv8/YOLOv11 using ZETIC Melange.
- [Speech Recognition (Whisper)](/tutorials/speech-recognition-whisper): Run OpenAI Whisper speech-to-text on-device with ZETIC Melange.
- [Tensor](/api-reference/android/Tensor): Complete API reference for the Tensor class on Android.
- [ZeticMLangeHFModel](/api-reference/android/ZeticMLangeHFModel): API reference for loading Hugging Face models on Android.
- [ZeticMLangeLLMModel](/api-reference/android/ZeticMLangeLLMModel): API reference for running LLM inference on Android.
- [ZeticMLangeModel](/api-reference/android/ZeticMLangeModel): API reference for general model inference on Android.
- [Enums and Constants](/api-reference/android/enums-and-constants): Android enum and support types for ZeticMLange.
- [RagPipeline](/api-reference/flutter/RagPipeline): API reference for retrieval-augmented generation in Flutter.
- [Tensor](/api-reference/flutter/Tensor): Complete API reference for the Tensor class in Flutter.
- [ZeticMLangeHFModel](/api-reference/flutter/ZeticMLangeHFModel): API reference for loading Hugging Face models in Flutter.
- [ZeticMLangeLLMModel](/api-reference/flutter/ZeticMLangeLLMModel): API reference for running LLM inference in Flutter.
- [ZeticMLangeModel](/api-reference/flutter/ZeticMLangeModel): API reference for general model inference in Flutter.
- [Enums and Constants](/api-reference/flutter/enums-and-constants): Flutter enum and support types for zetic_mlange.
- [Tensor](/api-reference/ios/Tensor): Complete API reference for the Tensor class on iOS.
- [ZeticMLangeHFModel](/api-reference/ios/ZeticMLangeHFModel): API reference for loading Hugging Face models on iOS.
- [ZeticMLangeLLMModel](/api-reference/ios/ZeticMLangeLLMModel): API reference for running LLM inference on iOS.
- [ZeticMLangeModel](/api-reference/ios/ZeticMLangeModel): API reference for general model inference on iOS.
- [Enums and Constants](/api-reference/ios/enums-and-constants): iOS enum and support types for ZeticMLange.
- [Advanced Configuration](/platform-integration/android/advanced-configuration): Advanced Melange configuration options for Android.
- [Basic Inference](/platform-integration/android/basic-inference): Run your first AI model inference on Android with ZETIC Melange.
- [Setup](/platform-integration/android/setup): Set up ZETIC Melange in your Android project.
- [Advanced Configuration](/platform-integration/flutter/advanced-configuration): Advanced Melange configuration options for Flutter.
- [Basic Inference](/platform-integration/flutter/basic-inference): Run your first AI model inference in Flutter with ZETIC Melange.
- [Setup](/platform-integration/flutter/setup): Set up ZETIC Melange in your Flutter project.
- [Advanced Configuration](/platform-integration/ios/advanced-configuration): Advanced Melange configuration options for iOS.
- [Basic Inference](/platform-integration/ios/basic-inference): Run your first AI model inference on iOS with ZETIC Melange.
- [Setup](/platform-integration/ios/setup): Set up ZETIC Melange in your Xcode project.
- [React Native](/platform-integration/react-native): ZETIC Melange React Native integration. Coming soon.

## Optional

- [Full documentation as plain text](/llms-full.txt): Complete documentation content for thorough context