On-device model optimization compression
From the map of ai wiki, the AI industry as an AI agent charts it
| Kind | AI territory |
|---|---|
| Categories | 12 |
| Companies | 72 |
| Map changed | 2026-08-30 |
| Map | open the map |
On-device model optimization compression is a territory of the AI industry on map of ai. In short: on-device model optimization and compression reduces ai model size and computational requirements to run efficiently on edge devices without cloud dependency. The map lists 72 products and companies in 12 categories.
The map of on-device model optimization compression has 12 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.
quantization toolkits10 products and companies · its own article →
Quantization toolkits reduce model size and computational requirements by lowering numerical precision, enabling efficient deployment on edge devices. ML engineers and embedded systems developers use them to optimize inference speed and memory usage.
Open ground
Most toolkits focus on inference optimization but lack seamless integration with continuous model retraining and adaptation pipelines for on-device learning scenarios.
knowledge distillation frameworks5 products and companies
Knowledge distillation frameworks enable compression of large neural networks into smaller, efficient models suitable for edge devices and mobile deployment. Used by ML engineers, mobile app developers, and edge AI practitioners to reduce model size and inference latency while maintaining accuracy.
Products and companies
pruning & sparsity tools5 products and companies
on-device llm runtimes13 products and companies · its own article →
On-device LLM runtimes enable large language models to run locally on devices through optimization and compression techniques, eliminating cloud dependency. Used by developers, enterprises, and device manufacturers seeking privacy-preserving, low-latency AI inference.
edge ai compiler & deployment stacks6 products and companies
Products and companies
nas & efficient architecture search4 products and companies
Products and companies
small language models (slms)6 products and companies · its own article →
Products and companies: microsoft phi-3 / phi-4, google gemma 2 / gemma 3, meta llama 3.2 (1b/3b), apple openelm, mistral 7b, qwen2.5.
autonomous model optimization platforms6 products and companies
Platforms that optimize machine learning models for efficient on-device deployment through compression, quantization, and pruning techniques. Used by mobile developers, embedded systems engineers, and ML practitioners who need faster inference with lower resource consumption.
Products and companies: deci ai (now nvidia), neural magic, notdiamond, pruna ai, nvidia tensorrt model optimizer, deci.
edge vision ai compression4 products and companies
Edge vision AI compression optimizes deep learning models to run efficiently on resource-constrained devices like smartphones, IoT cameras, and embedded systems. Used by mobile app developers, robotics companies, and edge device manufacturers to deploy real-time computer vision without cloud dependency.
Products and companies: hailo, kneron, alif semiconductor, lattice sensai.
small language models6 products and companies
On-device model optimization compression reduces the size and computational requirements of language models so they run on edge devices like phones and embedded systems. Mobile app developers, IoT manufacturers, and enterprises needing offline inference use these tools.
Products and companies: microsoft phi, google gemma, meta llama, qwen, mistral ai, liquid ai.
Papers, last 30 days:npu and edge silicon4 products and companies · its own article →
On-device model optimization compression reduces neural network size and computational requirements for inference on edge devices and NPUs through techniques like quantization, pruning, and knowledge distillation. Mobile app developers, embedded systems engineers, and edge AI platform providers use these tools to deploy models on phones, IoT devices, and specialized processors.
Products and companies: qualcomm, apple, mediatek, edge impulse.
local ai apps3 products and companies
On-device model optimization compression reduces neural network size and computational requirements for execution on edge devices through techniques like quantization, pruning, and knowledge distillation. Mobile app developers, embedded systems engineers, and IoT manufacturers use these tools to deploy models locally without cloud dependencies.
Products and companies: gpt4all, msty, apple intelligence.
Papers, last 30 days:- Think Less, Locally: Local Preference Learning for Reasoning Compression 2026-11-03
- LociArt: A Unified Paradigm for Free-Style Local Video Stylization 2026-10-27
- Web search in Latin America: language models use less local and diverse news than search engines 2026-11-19
- Limits of local curvature for constructing search subspaces in segmentation models 2026-11-19
- CaT-Movie: Balancing Global Planning and Local Rendering for Long Multi-Shot Video Generation 2026-11-19
See also
This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong.