# Local LLM Guide

> Practical guidance for running local language models on Mac, Windows, and Linux.

Start with hardware fit, choose one app, and follow a setup guide. Memory and cost tools calculate estimates from your inputs. A local app can also use cloud models or connected services; the selected workflow determines where data goes.

## Start here

- [Check your hardware](https://localllmguide.com/): Find a starting model size and app for your computer.
- [Choose a local AI app](https://localllmguide.com/software): Compare desktop apps, model runners, and document workspaces.
- [Local AI privacy](https://localllmguide.com/local-ai-privacy): Check model providers, document storage, and connected tools.

## Setup guides

- [What Is a Local LLM? Local AI Explained for Beginners](https://localllmguide.com/guides/what-is-local-ai): Learn what a local LLM is, how it differs from cloud AI, what hardware you need, and how to start with Ollama or LM Studio on your own computer.
- [How to Install Ollama on Mac, Windows, and Linux](https://localllmguide.com/guides/install-ollama): Install Ollama on Mac, Windows, or Linux, run your first local AI model, verify that it works, and avoid common storage, PATH, and first-run mistakes.
- [Download and Install LM Studio: Your First Local AI Model](https://localllmguide.com/guides/install-lm-studio): Download and install LM Studio on Mac, Windows, or Linux. Check hardware requirements, download a model that fits, start a chat, and fix common setup problems.
- [How to Install Open WebUI With Ollama](https://localllmguide.com/guides/open-webui-with-ollama): Set up Open WebUI with Ollama using Docker or Python, connect your local models, and fix common connection and missing-model errors.
- [Chat With PDFs Locally](https://localllmguide.com/guides/chat-with-pdfs-locally): Chat with PDFs locally using LM Studio, Open WebUI, or AnythingLLM. Set up document retrieval, check privacy settings, and fix inaccurate answers.
- [Local AI Glossary](https://localllmguide.com/guides/local-ai-glossary): Plain-English definitions for local AI terms, including local LLMs, model size, quantization, context length, VRAM, unified memory, embeddings, and RAG.
- [What Local LLM Can My Computer Run?](https://localllmguide.com/guides/what-local-llm-can-my-computer-run): Enter your hardware, understand the memory that matters, and choose a local model size that leaves enough headroom for context and normal work.
- [Local LLM RAM and VRAM Requirements](https://localllmguide.com/guides/local-llm-ram-vram-requirements): Understand how model weights, quantization, context, RAM, dedicated VRAM, and Apple unified memory determine which local LLMs are practical.
- [How to Run DeepSeek-R1 Locally](https://localllmguide.com/guides/run-deepseek-r1-locally): Choose the exact DeepSeek-R1 distilled model, check its memory requirements, and run it in Ollama or LM Studio with practical reasoning and licensing limits.
- [The Complete Guide to Local AI Coding: Setting Up Cline, Continue.dev, and Aider with Local Models](https://localllmguide.com/guides/local-ai-coding-agents): Connect Continue, Cline, and Aider to Ollama or LM Studio. Check model fit, context, edit quality, and network settings before using local coding assistants.
- [How to Use Model Context Protocol (MCP) with Local LLMs: LM Studio & Ollama Tutorial](https://localllmguide.com/guides/how-to-use-mcp-with-local-llms): Connect local LLMs to local files, databases, and APIs using the Model Context Protocol (MCP). Learn how to set up MCP in LM Studio and Ollama.
- [Run Local AI on Linux](https://localllmguide.com/guides/local-ai-linux-setup): Install Ollama on Linux, download a small local model, check the service and GPU, and fix common first-run failures without adding Docker.
- [Use Local AI Offline](https://localllmguide.com/guides/use-local-ai-offline): Prepare a local model for offline chat, writing, coding, or documents. Check downloads, local providers and a fresh offline restart before travelling.
- [Choose a Local AI Model for Coding](https://localllmguide.com/guides/local-models-for-coding): Choose a local coding model by memory and task, try Qwen2.5-Coder or a small Qwen3 model, and check generated code before connecting an agent.
- [Choose a Local AI Model for Writing](https://localllmguide.com/guides/local-models-for-writing): Use a local model for rewriting, outlines and drafting. Compare models with the same text, preserve facts and tone, and check memory and offline limits.
- [Run Local AI with 6GB of VRAM](https://localllmguide.com/guides/local-ai-6gb-vram): Start local AI on a 6GB graphics card with a small quantized model. Check GPU use, context memory and offloading before trying a larger download.
- [Run Local AI Without a GPU](https://localllmguide.com/guides/local-ai-cpu-only): Run local AI without a dedicated GPU using Ollama, LM Studio or llama.cpp. Choose a small model, preserve RAM headroom and test a realistic prompt.
- [Does Local AI Have Unlimited Tokens?](https://localllmguide.com/guides/local-ai-unlimited-tokens): Local models can avoid a hosted per-token charge, but context, response limits, memory, time and app licensing still matter. Learn which limit you are seeing.
- [Understand Local AI Context Windows](https://localllmguide.com/guides/local-ai-context-windows): Understand published and selected context limits, estimate text tokens, leave room for an answer, and check memory before increasing Ollama or LM Studio context.
- [Choose Model Compression for Local AI](https://localllmguide.com/guides/local-ai-quantization): Understand Q4, Q5 and Q8 model compression, choose a supported download, compare the same checkpoint, and avoid confusing file size with runtime memory.
- [Choose a GPU for Local AI](https://localllmguide.com/guides/choose-gpu-for-local-ai): Choose graphics hardware around the model and context you need. Check VRAM, app support, power and space, and compare real workloads before buying.
- [Move Local AI Models to Another Drive](https://localllmguide.com/guides/move-local-model-storage): Free space by moving Ollama or LM Studio model downloads. Copy the correct folders, change the model directory, and confirm a model works before deleting originals.

## Comparisons

- [Ollama vs LM Studio: Which Local AI App Should You Use?](https://localllmguide.com/compare/ollama-vs-lm-studio): Compare Ollama and LM Studio for local AI: setup, privacy, models, GUI, API support, Mac, Windows, document chat, and beginner use.
- [Best Local AI Setup for Mac](https://localllmguide.com/compare/best-local-ai-setup-for-mac): Set up local AI on a Mac with Ollama or LM Studio. See the best beginner stack, RAM requirements, model sizes, and Apple Silicon recommendations.
- [Run Local AI on Windows: App Choice and Setup](https://localllmguide.com/compare/best-local-ai-setup-for-windows): Learn the easiest way to run local AI on Windows with Ollama, LM Studio, or Open WebUI, including GPU, RAM, Docker, and storage tips.
- [Best Local AI for 8GB RAM](https://localllmguide.com/compare/best-local-ai-setup-for-8gb-ram): See what local AI models can realistically run on 8GB RAM, what to expect, and when to use Ollama, LM Studio, or a smaller model.
- [Best Local AI for 16GB RAM](https://localllmguide.com/compare/best-local-ai-setup-for-16gb-ram): Find the best local AI setup for a 16GB machine, including Ollama, LM Studio, Mac unified memory, Windows RAM, and VRAM limits.
- [Best Local AI for 32GB RAM](https://localllmguide.com/compare/best-local-ai-setup-for-32gb-ram): See what local AI models a 32GB computer can run, whether the upgrade is worth it, and the best Ollama or LM Studio setup.
- [GGUF vs MLX: Which Model Download Should You Choose?](https://localllmguide.com/compare/gguf-vs-mlx): Compare GGUF and MLX model downloads, check app and hardware support, preserve the checkpoint and license, and test an Apple Silicon setup fairly.
- [Ollama vs llama.cpp: Convenience or Direct Model Control?](https://localllmguide.com/compare/ollama-vs-llama-cpp): Choose Ollama for model management and app connections or llama.cpp for direct file and runtime control. Follow a small first run and compare the exact workload.
- [LM Studio vs Jan: Choose a Local AI Desktop App](https://localllmguide.com/compare/lm-studio-vs-jan): Compare LM Studio and Jan for desktop local AI. Choose a supported model, test the same task, and check optional cloud, web, tool and document settings.
- [Open WebUI vs AnythingLLM](https://localllmguide.com/compare/open-webui-vs-anythingllm): Compare Open WebUI and AnythingLLM for local AI chat, document workflows, setup, and data handling. Choose the workspace that fits your task.
- [Local AI vs Cloud AI](https://localllmguide.com/compare/local-ai-vs-cloud-ai): Compare local AI and cloud AI for privacy, hardware needs, running costs, and everyday tasks. Decide which parts of your workflow to run locally.

## Tools and reference

- [RAM and VRAM calculator](https://localllmguide.com/tools/ram-vram-calculator): Estimate memory fit from model size, context, and hardware inputs.
- [Local setup recommender](https://localllmguide.com/tools/stack-recommender): Choose a starting setup from your hardware and task.
- [App comparison tool](https://localllmguide.com/tools/compare): Compare features and supported workflows side by side.
- [Cloud vs local cost calculator](https://localllmguide.com/tools/cloud-vs-local-cost): Estimate running costs from your own usage and hardware assumptions.
- [Token and PDF counter](https://localllmguide.com/tools/token-counter): Count text with a supported model tokenizer and leave room for instructions and a reply.
- [Browser AI chat](https://localllmguide.com/tools/browser-ai): Explicitly download a small model and try temporary chat on a supported device.
- [Local model shortlist](https://localllmguide.com/models): Browse models by hardware and task.
- [Hardware profiles](https://localllmguide.com/hardware): Understand memory constraints and starting model sizes.
- [Troubleshooting](https://localllmguide.com/troubleshooting): Find the next check for a model, GPU, connection, or PDF problem.
- [MCP server directory](https://localllmguide.com/mcp/servers): Compare server tasks, connections, and permissions.

## Optional

- [Local AI news](https://localllmguide.com/news): Published updates about apps, models, and local workflows.
- [News RSS feed](https://localllmguide.com/feed.xml): Subscribe to published articles with their original publication dates.
- [About Local LLM Guide](https://localllmguide.com/about): Site purpose and editorial responsibility.
- [Methodology](https://localllmguide.com/methodology): How the site approaches recommendations and calculations.