On Device Model Inference

Foundry Local runs AI models on-device using ONNX Runtime (CPU/GPU/NPU) to keep data local, offering an OpenAI-compatible API, Python/JS/C#/Rust SDKs, a model hub, and CLI tools for edge and enterprise deployments.

Sentiance processes sensor data on-device to generate real‑time behavioral insights for drivers and mobile users, enabling safety monitoring, fraud detection, usage‑based insurance, and personalized in‑vehicle features while keeping data privacy and bandwidth minimal.

gpt-oss playground provides open-weight demos of gpt-oss-120b and 20b for infrastructure testing, distributed and on-device inference, benchmarking, API integration, and reproducible research, with adjustable reasoning levels and visible-reasoning for diagnostics. Demo-only; validate outputs.

LLMWare AI installs a lightweight client on PCs, providing instant access to 100+ AI models optimized for Intel and Qualcomm hardware. It supports RAG, auto‑tunes weights, runs locally without Wi‑Fi, and offers an admin console for monitoring, scaling, and audit logs.

Inferless is a serverless platform for deploying machine learning models seamlessly. It offers automatic load balancing, custom runtime environments, and automated CI/CD workflows, minimizing infrastructure management while scaling efficiently from single to millions of requests.

ZETIC deploys TorchScript, TensorFlow, and ONNX models to mobile and embedded devices, quantizing for CPU, GPU, or NPU to reach up to 60× speed and 50% size reduction. It supplies benchmarks and a 3‑line offline code snippet for privacy‑preserving AI.

Nexa AI offers an on‑device platform that lets developers deploy vision, audio, and text models to NPUs, GPUs, and CPUs with one line of code. The SDK supports day‑zero deployment, multimodal inference, and optimizations for mobile, automotive, and IoT devices.

Pioneer automates retraining and deployment of open-source models, using live inference data for fine-tuning and one-shot adaptation. It manages adaptive inference, routing, RAG pipelines, agent workflows, synthetic data generation, monitoring, and automated checkpoint promotion.

openmed is an on-device clinical AI platform for PHI/PII detection, de-identification, and healthcare NER, offering 1,000+ model variants, multilingual support, curated biomedical datasets, configurable privacy controls, and air-gapped macOS/iOS/server runtimes.

Teachable Machine lets users build TensorFlow.js image, audio, or pose classifiers through a browser interface without coding. Collect, label, and train data in‑browser, evaluate accuracy, then export models to web, Node.js, Coral, or Arduino for rapid prototyping and educational projects.

canirun.ai is a searchable database mapping AI models to compatible hardware, listing CPUs/GPUs (including Apple M-series and NVIDIA cards), model requirements, VRAM/memory needs, filters and comparisons to plan local inference, fine-tuning, or deployment.

On-Device AI is a local-run assistant for Apple devices, offering offline chat, document searches, and image analysis. It integrates with Siri and provides customizable settings, to-do lists, and reminders for enhanced productivity and data privacy.

Modal is a cloud‑native platform that lets developers run inference, training, batch jobs, sandboxes, and notebooks with sub‑second cold starts and instant autoscaling. It’s Python‑centric, offers elastic multi‑cloud GPU scaling, zero‑idle scaling, unified observability, and high‑throughput AI‑nativ

OpenRouter gives one API key to access 300+ models from 60+ providers, SDK‑compatible, with visual routing, automated fall‑back, edge hosting, data‑policy controls, and agentic tools for building efficient autonomous workflows.

fal.ai offers a unified API for generating images, videos, audio, and 3D models from a library of over 1,000 production‑ready assets. It provides serverless GPU inference, private deployment options, NVIDIA‑cluster fine‑tuning, SOC 2 compliance, and enterprise‑grade support.

Wafer AI is a serverless inference platform that lets you run open-source LLMs in production with OpenAI-compatible APIs. It offers dedicated endpoints with optimized performance, long-context support, and caching to reduce costs for coding, reasoning, and agent workloads.

AI-driven IoT device management platform that automates discovery, inventory and secure onboarding, offers real-time monitoring, visual analytics, ML-based anomaly detection and predictive maintenance, role-based security, APIs for integrations, mobile/web access, alerts and exportable reports.

Imagen is a generative AI model by Google DeepMind that produces high-quality, photorealistic images from natural language prompts using advanced diffusion techniques. It supports creative applications in design, media, and content generation.

Venice is a private on‑device AI platform that offers a broad array of open‑source models for text, image, code, and character creation. It includes a chat interface, an agent‑building API, watermark removal, upscaling, and batch image generation.

Inception Labs' diffusion-based large language models (dLLMs) offer faster, more efficient, and cost-effective text generation than traditional autoregressive models. With built-in error correction, multimodal support, and structured output control, they excel in function calling and complex data ge

SiliconFlow is an AI infrastructure platform enabling high-speed inference for LLMs and multimodal applications, supporting serverless, reserved, and private-cloud deployments. It offers low-latency processing, elastic compute, and built-in monitoring for scalable, cost-efficient AI workloads.

Roboflow streamlines computer‑vision projects by offering a low‑code pipeline for data annotation, GPU‑accelerated training, and multi‑environment deployment. It integrates with PyTorch, TensorFlow, Hugging Face, major clouds, and meets SOC2 Type 2 and HIPAA security.

Compact edge platform featuring the Hailo‑8 accelerator for up to 83 TOPs. Supports USB, PCIe, Ethernet, and GPIO; runs Linux ≥ 6.18 with drivers, enabling rapid AI deployment for real‑time inference in automotive, security, and industrial inspection.

Groq is an inference platform that uses custom LPU silicon for low‑latency, high‑throughput AI workloads. It supports large language and multimodal models via an OpenAI‑compatible API, with modular deployment and predictable performance for NLP, vision, and recommendation tasks.

Use.ai is an AI Workspace platform unifying access to over 25 AI models including ChatGPT, Claude, and Gemini, offering a single interface for versatile AI applications and seamless model switching.

Liner.ai is a cross‑platform no‑code ML app that trains models locally in minutes on images, text, audio, or video. It auto‑selects algorithms, offers ready‑to‑use templates, and exports models for web, mobile, or edge deployment.

local.ai runs language models locally without GPUs. Its Rust backend keeps the binary under 10 MB and performs CPU inference with GGML quantization. A single‑click interface streams responses to a UI, while a model manager tracks, verifies, and resumes downloads.

Cerebras provides a wafer-scale AI accelerator and software stack that enables single-node training of very large LLMs, high-throughput low-latency inference (GLM-4.6 at 1,000 TPS), PyTorch SDK, deployment options, and MLOps tooling.

Draw Things is an AI-assisted image-generation app for iOS, iPadOS, and macOS that converts text prompts into images, supports on-device model downloads and LoRA training, image-to-video conversion, infinite canvas, style transfer, and model import/export.

VModel provides a unified REST API that lets developers deploy and run custom or community‑built models with a single line of code. It supports Node.js, Python, and cURL for image, text, and video tasks, automatically scaling for production workloads.

LM Studio runs open‑source large language models locally on Mac (M‑series), Windows, and Linux, enabling private, offline inference. It offers command‑line and headless deployment, server‑side API, SDKs, a model hub, and LM Link for remote model access.

devAIce® extracts over 7,000 acoustic parameters via its SDK, Web API, and Unity/Unreal plug‑ins, delivering real‑time voice‑expression analytics for XR, automotive, robotics, and healthcare. It supports stress and health biomarker detection, emotion‑aware interfaces, and GDPR‑compliant data handlin

Enclave AI runs large language models on Mac and iPhone, keeping all text, voice, and document processing offline. It supports on‑device speech recognition, synthesis, and custom assistants, with local, encrypted conversation history and PDF summarization, ensuring privacy.

Cloud Phone is a cloud-based platform that enables users to run real Android environments, manage multiple devices concurrently, and automate applications securely, offering customizable profiles and insights for effective marketing and account management.

UBIAI fine‑tunes LLMs with classifiers, retrievers, and reasoning. It automates PDF/DOCX labeling, synthetic data, and quality filtering; offers 15‑minute prompt‑level tuning or 2‑4 hour weight training; exports to GGUF, safetensors, or Hugging Face for API or custom deployment.

DeepSense.ai provides end‑to‑end AI solutions for enterprises, integrating large language models, retrieval‑augmented generation, MLOps, advanced computer‑vision, edge inference, and predictive analytics to deliver scalable, real‑time AI agents, co‑pilots, and maintenance optimization.

Pi.dev is a terminal-based coding agent for developers and teams, offering TUI, RPC/SDK modes, extensible TypeScript modules and packages, multi-provider model authentication, session/context management, RAG/memory primitives, and CLI tooling for automation and integration.

ModelsLab offers API‑based generative AI for image, video, audio, and language tasks, including editing, generation, and voice synthesis. It supports GPU server deployment, custom workflows, fine‑tuning, and LoRA adaptation for creators and developers.

EmpirioLabs AI is a platform for hosting, deploying, and scaling open-source and proprietary AI models via API or web playground. It supports multimodal, long-context models with optimized endpoints, creative templates, and high-throughput rate limits for production workloads.

Runpod supplies on‑demand GPUs in 31 regions, offering single‑node pods, multi‑node clusters, and serverless workloads. It delivers low‑latency inference, efficient fine‑tuning, instant scaling, S3‑compatible storage, real‑time logs, and sub‑200 ms cold starts.

InsightAI delivers AI‑driven fraud and AML intelligence, using device fingerprints, network signals, and behavioral analytics to detect fraud before transactions, automate case summarization, spot forged documents, and provide millisecond‑level real‑time risk scoring with explainable outputs for aud

H2O.ai delivers an end‑to‑end AI platform that automates feature engineering, model selection, and explainability through AutoML, offers no‑code LLM training, supports enterprise multi‑model orchestration, and includes MLOps and a feature store, all compliant with strict data security standards.

DeviceHub is an IoT device management platform that enables efficient deployment, proactive monitoring, and intelligent automation. It enhances device performance and decision-making through AI-driven analytics, streamlining operations for businesses managing connected hardware.

Ultralytics offers a platform for developing and deploying visual AI solutions across industries, utilizing YOLO for advanced data analysis and object detection. Its user-friendly interface aids in efficient training and deployment of machine learning models.

AI Fiesta lets you run multiple AI models side-by-side in one chat with preserved context, automated model selection, prompt enhancement, image generation, audio transcription, expert avatars and project-wide modes for consistent content, research, and code review workflows.

PicFinder.AI (Runware) is an AI image-generation platform with 300,000+ models and style adapters, offering batch generation, infinite-scroll outputs, fine-grained controls and scalable, optimized rendering via a centralized Runware Playground for experimentation and production workflows.

Lightning AI is a PyTorch Lightning‑based cloud platform for training, deploying, and serving models at scale. It offers GPU workspaces, managed clusters, fractional pay‑as‑you‑go GPU capacity, inference APIs, serverless deployment, security, and integration with LitServe, LitGPT, and LLMs.

Halo is an open‑source AR glasses platform with OLED display, bone‑conduction audio, and on‑device AI powered by Alif B1 Cortex‑M55, enabling real‑time multimodal conversations, context capture, and cross‑platform app development via Lua on ZephyrOS.

Img2Prompt is an AI tool that generates text prompts from images and provides a public API for image captioning and prompt-based image generation. It can run on personal hardware or cloud platforms.

Graphite Note is a user-friendly, no-code predictive analytics tool for cross-industry teams. It delivers accurate predictions (outcomes, lead conversions), analyzes customer behavior, creates personalized marketing strategies, optimizes campaigns, and forecasts demand, simplifying complex data ana