Guides
Find the right AI guide faster.
Guides are step-by-step walkthroughs for building with open-source AI — each one takes you from concept to a working setup. Search guides covering local AI, RAG, coding agents, MCP, security, deployment, audio transcription, model selection, and workflow planning.
Updated as tools, models, and setups change
Practical over promotional
Guides prioritize working setup steps over generic AI hype.
Sourced and testable
Steps link to official docs, tools, and models you can verify yourself.
Updated as tools change
Guides are revisited as runtimes, models, and setups evolve.
Start here
Start with a practical build path
These guides answer the most common first questions before you choose tools, models, or integrations.
Local AI
Set up local LLMs on Windows
Start here when you want Ollama, LM Studio, Open WebUI, and model testing working on a Windows machine.
Read setup guide →RAG
Build a private local RAG stack
Move from plain chat to document retrieval with Ollama, Open WebUI, Qdrant, embeddings, and source checks.
Build RAG →Models
Choose the right model for the job
Match coding, RAG, summarization, agents, and media workflows to model families and hardware limits.
Choose models →Knowledge base tracks
Browse guides by technical track
Select a track to focus on that topic, or browse all tracks below. Each track maps to a category in the OpenSourcesAI Knowledge Base.
Foundations
Core concepts before you choose tools, models, or hardware.5 guidesA plain-English guide to local large language models: how they work, what hardware they need, when to use them, and how to get started with Ollama or LM Studio.
Read guide →FoundationsChoose an AI Model by Use CaseA builder-first decision guide for choosing open models by task instead of hype.
Read guide →FoundationsOpen-Source AI Stack for Small BusinessesA practical AI stack for small teams that need privacy, automation, and useful internal tools.
Read guide →FoundationsOpen Weight vs Open Source AIWhat "open source AI" actually means under the OSI definition, how it differs from open weights and openly licensed weights, and where every model in this catalog stands.
Read guide →FoundationsHow to Evaluate Local Models Before ProductionA practical evaluation process for local and open-weight models before real users depend on them.
Read guide →Hardware & VRAM
GPU memory, quantization tradeoffs, and setup walkthroughs.5 guidesUnderstand GPU memory limits, quantization tradeoffs, KV cache math, and the exact formulas to match model size to your hardware before downloading anything.
Read guide →Hardware & VRAMWhat Is Quantization?Compressing model weight precision (e.g. 16-bit → 4-bit) to cut memory use. Q4 models run on half the VRAM with a small quality tradeoff.
Read guide →Hardware & VRAMHow to Set Up Local LLMs on Windows in 2026A practical Windows setup guide for Ollama, LM Studio, Open WebUI, coding assistants, model testing, and local RAG foundations.
Read guide →Hardware & VRAMLinux for Local AI: What Actually Breaks and How to Fix ItA calm, source-backed Linux setup guide: NVIDIA Secure Boot and driver updates, AMD ROCm vs Vulkan, RTX 50 requirements, kernel age, Intel Arc, and verification commands.
Read guide →Hardware & VRAMHow Much VRAM Do You Need for Local AI?A practical explanation of VRAM, quantization, context length, and model size for local AI builders.
Read guide →RAG & Data
Retrieval pipelines, vector databases, and document ingestion.8 guidesThe set of separate programs you run together to get working local AI: model, runtime, interface, and optional retrieval and tools. Which layers you need, and the smallest stack that works.
Read guide →RAG & DataWhat Is RAG?Retrieval-augmented generation in plain English: how a model answers from your own documents by looking them up at question time, why it is not fine-tuning, and when you do not need it.
Read guide →RAG & DataBuild a Local RAG Stack with Ollama, Open WebUI, and QdrantA practical local RAG guide for document prep, Qdrant, embeddings, retrieval testing, Open WebUI, and Ollama workflows.
Read guide →RAG & DataHow to Choose a Vector Database for a Small RAG ProjectA practical guide to choosing between Chroma, Qdrant, pgvector, Weaviate, Milvus, and LanceDB for small RAG apps.
Read guide →RAG & DataWhat Is a Reranker in RAG?A practical explanation of rerankers, where they fit after vector search, and how to evaluate them in RAG pipelines.
Read guide →RAG & DataHow to Build a Private AI Chatbot with Local ModelsPlan a private chatbot stack using local models, retrieval, permissions, and self-hosted interfaces.
Read guide →RAG & DataAPI Key Security for AI BuildersA practical security checklist for AI builders managing model provider keys, vector database credentials, hosting access, and team vaults.
Read guide →RAG & DataWeb Data for AI Apps: Responsible Workflows for RAG, Research, and MonitoringA practical guide to public web data workflows for AI apps, RAG enrichment, market intelligence, SERP monitoring, and research automation.
Read guide →IDE Workflows
AI coding assistants, open-weight coding models, and tooling.2 guidesA practical testing guide for Qwen, DeepSeek, Kimi, GLM, and other open-weight coding model families.
Read guide →IDE WorkflowsOllama vs LM Studio vs JanA three-way guide to choosing a local model runner for CLI, desktop, and open-source workflows.
Read guide →Agents & MCP
Model Context Protocol, tool-use, and agent security patterns.3 guidesA practical explanation of Model Context Protocol, reusable MCP servers, and where MCP fits in AI agent workflows.
Read guide →Agents & MCPPractical AI Agent Stack Using Open-Source ToolsBuild an agent stack with models, tools, memory, workflows, tracing, and human review.
Read guide →Agents & MCPRun Grok Build with a Local Model (Ollama)Point xAI's Apache 2.0 terminal coding agent at a model on your own GPU through Ollama's OpenAI-compatible endpoint, with the config.toml recipe and a clear line between what xAI documents and what is untested.
Read guide →Audio & Media
Speech-to-text, Whisper model sizes, and transcription workflows.1 guideShowing 24 of 24 guides.
Guide
What Is a Local LLM?
A plain-English guide to local large language models: how they work, what hardware they need, when to use them, and how to get started with Ollama or LM Studio.
Read guide →Guide
What Is VRAM and How Much Do You Need for Local AI?
Understand GPU memory limits, quantization tradeoffs, KV cache math, and the exact formulas to match model size to your hardware before downloading anything.
Read guide →Guide
What Is an AI Stack?
The set of separate programs you run together to get working local AI: model, runtime, interface, and optional retrieval and tools. Which layers you need, and the smallest stack that works.
Read guide →Guide
What Is RAG?
Retrieval-augmented generation in plain English: how a model answers from your own documents by looking them up at question time, why it is not fine-tuning, and when you do not need it.
Read guide →Guide
How to Set Up Local LLMs on Windows in 2026
A practical Windows setup guide for Ollama, LM Studio, Open WebUI, coding assistants, model testing, and local RAG foundations.
Read guide →Guide
Build a Local RAG Stack with Ollama, Open WebUI, and Qdrant
A practical local RAG guide for document prep, Qdrant, embeddings, retrieval testing, Open WebUI, and Ollama workflows.
Read guide →Guide
How to Choose a Vector Database for a Small RAG Project
A practical guide to choosing between Chroma, Qdrant, pgvector, Weaviate, Milvus, and LanceDB for small RAG apps.
Read guide →Guide
How to Test Open-Weight Coding Models in 2026
A practical testing guide for Qwen, DeepSeek, Kimi, GLM, and other open-weight coding model families.
Read guide →Guide
Whisper Model Sizes: Tiny vs Base vs Small vs Medium vs Large
Compare Whisper tiny, base, small, medium, large, large-v3, large-v3-turbo, and faster-whisper for local transcription workflows.
Read guide →Guide
What Is MCP for AI Agents?
A practical explanation of Model Context Protocol, reusable MCP servers, and where MCP fits in AI agent workflows.
Read guide →Guide
Choose an AI Model by Use Case
A builder-first decision guide for choosing open models by task instead of hype.
Read guide →Guide
Open Weight vs Open Source AI
What "open source AI" actually means under the OSI definition, how it differs from open weights and openly licensed weights, and where every model in this catalog stands.
Read guide →Guide
What Is Quantization?
Compressing model weight precision (e.g. 16-bit → 4-bit) to cut memory use. Q4 models run on half the VRAM with a small quality tradeoff.
Read guide →Guide
Linux for Local AI: What Actually Breaks and How to Fix It
A calm, source-backed Linux setup guide: NVIDIA Secure Boot and driver updates, AMD ROCm vs Vulkan, RTX 50 requirements, kernel age, Intel Arc, and verification commands.
Read guide →Guide
What Is a Reranker in RAG?
A practical explanation of rerankers, where they fit after vector search, and how to evaluate them in RAG pipelines.
Read guide →Guide
Ollama vs LM Studio vs Jan
A three-way guide to choosing a local model runner for CLI, desktop, and open-source workflows.
Read guide →Guide
How to Build a Private AI Chatbot with Local Models
Plan a private chatbot stack using local models, retrieval, permissions, and self-hosted interfaces.
Read guide →Guide
How Much VRAM Do You Need for Local AI?
A practical explanation of VRAM, quantization, context length, and model size for local AI builders.
Read guide →Guide
Open-Source AI Stack for Small Businesses
A practical AI stack for small teams that need privacy, automation, and useful internal tools.
Read guide →Guide
How to Evaluate Local Models Before Production
A practical evaluation process for local and open-weight models before real users depend on them.
Read guide →Guide
Practical AI Agent Stack Using Open-Source Tools
Build an agent stack with models, tools, memory, workflows, tracing, and human review.
Read guide →Guide
API Key Security for AI Builders
A practical security checklist for AI builders managing model provider keys, vector database credentials, hosting access, and team vaults.
Read guide →Guide
Web Data for AI Apps: Responsible Workflows for RAG, Research, and Monitoring
A practical guide to public web data workflows for AI apps, RAG enrichment, market intelligence, SERP monitoring, and research automation.
Read guide →Guide
Run Grok Build with a Local Model (Ollama)
Point xAI's Apache 2.0 terminal coding agent at a model on your own GPU through Ollama's OpenAI-compatible endpoint, with the config.toml recipe and a clear line between what xAI documents and what is untested.
Read guide →Hands-on next step
Use guides with the Playground and stack recipes.
Read the guide, test the idea in browser when possible, then move into the relevant stack page to build the workflow.
For builders
Built an AI tool or open-source project?
Submit it for review or sponsor a featured placement on OpenSourcesAI. For sponsorship options, advertise with us. For submissions or corrections, use the submit page.