AI / ML Tools — Leveled Learning Map

Loading…

The idea: tools sorted by concept depth — how much you need to understand to really use each one — from Level 0 (use ready-made AI, no theory) to Level 4 (research- and production-scale systems). It's not about how hard a tool is to click; ChatGPT is easy to use (Level 0) even though the ideas behind it are advanced.

Every tool has two links: 📖 Read — an approachable article/tutorial to understand it — and 🐙 Code — its repo or homepage. Levels are judgment calls; you might move a tool ±1 level, and that's fine.

All levels L0 L1 L2 L3 L4

Awesome AI/ML Tools — 211 tools by concept depth (Level 0 to 4)An awesome list of AI and machine-learning tools for LLMs & Generative AI — searchable and leveled by concept depth (Level 0 to 4). Every tool links to an approachable article to read and to its code or homepage. Enable JavaScript for search, autocomplete and level/category filters.Level 0 — Consume ready-made GenAIUse finished AI products. No code, no theory.Chat assistants

ChatGPT — Read: How to use ChatGPT: A beginner's guide · Website

Claude — Read: Getting started with Claude.ai · Website

Google Gemini — Read: How to use Gemini: A beginner's guide · Website

Perplexity AI (AI answer engine) — Read: Getting started with Perplexity · Website

DeepSeek (free reasoning chat) — Read: A Complete How-To Guide to DeepSeek Chatbot · Website

Grok (real-time, X-integrated) — Read: How to Use Grok: A Beginner's Guide · Website

Microsoft Copilot (free, Windows-integrated) — Read: Copilot tutorial: Start using Copilot · Website

Coding assistants

GitHub Copilot — Read: GitHub for Beginners: How to get started with GitHub Copilot · Website

Cursor — Read: How To Use Cursor AI: A Complete Guide · Website

Windsurf — Read: Windsurf AI Agentic Code Editor: Features, Setup, Use Cases · Website

Claude Code (agentic coding CLI) — Read: Claude Code — Overview · Website

Media generation

ElevenLabs (voice / TTS) — Read: Text to Speech (product guide) · Website

Runway (AI video) — Read: Getting Started with Generative Video · Website

Suno (AI music) — Read: Suno Help Center — Getting started · Website

Leonardo.Ai (AI image) — Read: How to Use Leonardo AI · Website

Research assistants

NotebookLM (source-grounded research) — Read: Introducing NotebookLM (Google) · Website

Level 1 — Run & build with no/low codeRun models locally via a GUI, or assemble apps by drag-and-drop. First ideas: prompts, tokens, RAG.Local model runtimes

Ollama — Read: How to Use Ollama to Run LLMs Locally · Code

LM Studio — Read: LM Studio Tutorial: Get Started with Local LLMs · Website

Jan — Read: Jan Desktop QuickStart · Website

GPT4All — Read: GPT4All Documentation · Code

Open WebUI — Read: Open WebUI Documentation · Code

LibreChat — Read: Quick Start — LibreChat · Website

LocalAI (self-hosted OpenAI API) — Read: Quickstart — LocalAI · Code

KoboldCpp (local model UI) — Read: How to Use KoboldCpp for AI Story Writing · Code

SillyTavern — Read: Quick Start — SillyTavern · Code

LobeChat (self-host chat UI) — Read: LobeChat installation guide · Code

Cherry Studio (desktop LLM workbench) — Read: Cherry Studio — README · Code

No-code / low-code builders

Flowise — Read: How to Build No-Code AI Agents Using Flowise AI · Code

Langflow — Read: Langflow: A Guide With Demo Project · Code

Dify — Read: Dify: A Guide With Demo Project · Code

n8n (workflow automation) — Read: Tutorial: Build an AI workflow in n8n · Website

LLMStack — Read: Introducing LLMStack · Code

AnythingLLM — Read: AnythingLLM: Complete Guide to Setup, RAG, and Use Cases · Code

Quivr (personal RAG) — Read: Quivr: Your Personal, Conversational Knowledge Base · Code

Lovable (prompt-to-app) — Read: Lovable — Getting started · Website

bolt.new (in-browser app builder) — Read: Get to know Bolt · Code

Coze (no-code agent builder) — Read: Coze Quickstart · Code

Onyx (enterprise RAG assistant) — Read: Onyx Documentation · Code

Access & studios

OpenRouter (model router) — Read: OpenRouter Quickstart · Website

Google AI Studio — Read: Google AI Studio Tutorial · Website

aider (terminal pair programmer) — Read: Aider: Terminal-Based AI Pair Programming Guide · Website

Open Interpreter — Read: Introduction — Open Interpreter · Code

Cline (OSS coding agent) — Read: Cline for new coders · Code

Smithery (MCP server marketplace) — Read: Smithery — MCP registry · Code

Level 2 — Code with high-level librariesWrite Python against SDKs and batteries-included frameworks; embeddings, basic orchestration, demos.Model provider SDKs / APIs

OpenAI API — Read: Developer quickstart — OpenAI API · Website

Anthropic Claude API — Read: Get started with Claude · Code

Cohere — Read: Cohere API Tutorial: Getting Started · Website

Mistral AI — Read: Quickstarts — Mistral AI · Website

Together AI — Read: Quickstart — Together AI · Website

Groq (fast inference) — Read: Quickstart — GroqDocs · Website

DeepSeek API (V3/R1, OpenAI-compatible) — Read: Your First API Call — DeepSeek · Code

xAI Grok API (Grok models) — Read: xAI Docs — Overview · Code

Google Gen AI SDK (unified Gemini SDK) — Read: Google Gen AI SDK — docs · Code

Fireworks AI (fast serverless inference) — Read: Serverless Quickstart — Fireworks · Code

Cerebras Inference (ultra-fast inference) — Read: Cerebras Inference Quickstart · Code

SambaNova Cloud (fast open-model inference) — Read: SambaNova Cloud Quickstart · Website

Orchestration frameworks

LangChain — Read: A Beginner's Guide to LangChain · Code

LlamaIndex — Read: Starter Tutorial (Using OpenAI) · Code

Haystack — Read: Get Started (Quickstart) — Haystack · Code

LiteLLM (unified API proxy) — Read: LiteLLM: A Guide With Practical Examples · Code

txtai (embeddings DB) — Read: Introducing txtai, the all-in-one embeddings database · Code

LLMWare (RAG + small models) — Read: LLMWare Documentation · Code

Vercel AI SDK (TypeScript AI toolkit) — Read: AI SDK — Introduction · Code

mem0 (agent memory layer) — Read: mem0 Documentation · Code

Models & embeddings

Hugging Face Transformers — Read: Transformers Quickstart (Quicktour) · Code

sentence-transformers (SBERT embeddings) — Read: Quickstart — Sentence Transformers · Code

App / demo UIs

Gradio — Read: Gradio Quickstart · Code

Streamlit — Read: Get started with Streamlit · Code

Chainlit — Read: Get Started: In Pure Python · Code

Structured output

Guidance — Read: Introduction to guidance · Code

Outlines — Read: Welcome to Outlines (docs intro) · Code

LMQL (LLM query language) — Read: LMQL — Getting Started · Code

Instructor (Pydantic-validated outputs) — Read: Instructor — Structured LLM Outputs · Code

BAML (typed prompt DSL) — Read: BAML Documentation · Code

Protocols & starter agents

MCP (tool-connect protocol) — Read: Model Context Protocol Explained in 3 Levels of Difficulty · Code

CrewAI — Read: CrewAI Quickstart · Code

smolagents — Read: smolagents — Guided tour · Code

GPT Researcher (research agent) — Read: GPT Researcher — Introduction · Code

Pydantic AI — Read: Pydantic AI: A Beginner's Guide · Code

Langroid (multi-agent) — Read: Langroid: Multi-Agent Programming (docs) · Code

OpenAI Agents SDK (official multi-agent SDK) — Read: OpenAI Agents SDK — docs · Code

FastMCP (build MCP servers) — Read: FastMCP — Welcome · Code

MCP Registry (MCP server registry) — Read: Introducing the MCP Registry · Code

Level 3 — Build, fine-tune & shipEngineer RAG/agent systems, fine-tune models, use vector DBs, evaluate, and operate (LLMOps, serving).Advanced agent frameworks

LangGraph — Read: LangGraph Tutorial: What Is LangGraph and How to Use It? · Code

AutoGen / AG2 — Read: AgentChat Quickstart — AutoGen · Code

MetaGPT — Read: Quickstart (MetaGPT) · Code

CAMEL — Read: Building a Multi-Agent System with CAMEL AI · Code

Semantic Kernel — Read: How to quickly start with Semantic Kernel · Code

AutoGPT — Read: AutoGPT Guide: Creating Autonomous AI Agents · Code

OpenHands (AI software developer) — Read: OpenHands: Open Source AI Software Developer · Code

browser-use (browser automation) — Read: Quickstart — Browser Use · Code

Google ADK (Agent Development Kit) — Read: Python Quickstart — Agent Development Kit · Code

Mastra (TypeScript agents) — Read: Quickstart — Mastra · Code

DSPy (prompt optimization) — Read: Program, don't prompt (DSPy) · Code

Agno (high-perf agents (ex-Phidata)) — Read: Welcome to Agno · Code

Letta (stateful memory agents) — Read: Letta — Overview · Code

Stagehand (AI browser automation) — Read: Stagehand Documentation · Code

RAG frameworks & ingestion

RAGFlow — Read: Introduction to RAGFlow · Code

GraphRAG — Read: Getting Started — GraphRAG · Code

R2R (RAG server) — Read: What is R2R (official intro) · Code

AutoRAG (auto-optimize RAG) — Read: Tutorial — AutoRAG · Code

Neo4j (graph DB for KG-RAG) — Read: Get started with Neo4j · Website

Firecrawl — Read: Firecrawl: AI Web Crawler for LLM Apps · Code

Crawl4AI — Read: Quick Start — Crawl4AI · Code

Docling (document parsing) — Read: Quickstart — Docling · Code

Unstructured (document ingestion) — Read: Overview — Unstructured open source · Code

MinerU (PDF/DOCX to Markdown OCR) — Read: MinerU Documentation · Code

Marker (PDF to Markdown) — Read: Marker — README · Code

MarkItDown (file to Markdown) — Read: MarkItDown — README · Code

LightRAG (lightweight graph RAG) — Read: LightRAG — README · Code

Vector databases

Chroma — Read: Embeddings and Vector Databases With ChromaDB · Code

Pinecone — Read: Pinecone Quickstart · Website

Weaviate — Read: Weaviate Quickstart · Code

Qdrant — Read: Local Quickstart — Qdrant · Code

Milvus — Read: Milvus Vector Database Quickstart (Zilliz) · Code

LanceDB — Read: LanceDB Quickstart · Code

pgvector (Postgres vectors) — Read: pgvector Tutorial: Vector Search in PostgreSQL · Code

FAISS — Read: Introduction to Facebook AI Similarity Search (Faiss) · Code

Deep Lake (vector / data lake) — Read: What is Activeloop Deep Lake? Features & Getting Started · Code

Vespa (vector + text at scale) — Read: Vespa.ai · Code

Fine-tuning

PEFT (LoRA/QLoRA) — Read: LoRA (PEFT conceptual guide) · Code

TRL (RLHF / RL for LLMs) — Read: TRL — Transformer Reinforcement Learning · Code

Unsloth — Read: Fine-tuning LLMs Guide — Unsloth · Code

Axolotl — Read: Axolotl Getting Started · Code

LLaMA-Factory — Read: LLaMA-Factory WebUI Beginner's Guide · Code

torchtune — Read: torchtune Overview · Code

LitGPT — Read: Zero to LitGPT: Getting Started · Code

bitsandbytes (quantization) — Read: bitsandbytes (Transformers quantization guide) · Code

ms-swift (600+ LLM fine-tune/deploy) — Read: ms-swift Documentation · Code

mergekit (model merging) — Read: mergekit — README · Code

LLM Compressor (GPTQ/AWQ quantization) — Read: LLM Compressor — README · Code

Evaluation

Ragas (RAG evaluation) — Read: Quick Start — Ragas · Code

DeepEval — Read: DeepEval 5-min Quickstart · Code

promptfoo — Read: Promptfoo Tutorial: Hands-On LLM Evaluation · Code

Giskard (testing / red-team) — Read: Giskard — AI Evaluation & Red Teaming (docs) · Code

OpenAI Evals — Read: Evals — OpenAI Developers · Code

OpenCompass — Read: Quick Start — OpenCompass · Code

Inspect AI (UK AISI eval framework) — Read: Inspect AI Documentation · Code

Guardrails & security

Guardrails AI — Read: In-Application Quickstart — Guardrails AI · Code

NeMo Guardrails — Read: Introduction — NVIDIA NeMo Guardrails · Code

LLM Guard — Read: Get Started — LLM Guard · Code

PurpleLlama (Meta LLM safety) — Read: Announcing Purple Llama · Code

garak (LLM red-teaming scanner) — Read: garak — LLM vulnerability scanner · Code

Observability & LLMOps

Langfuse — Read: Get Started — Langfuse · Code

LangSmith — Read: Tracing quickstart — LangSmith · Website

Arize Phoenix — Read: What is Arize Phoenix? · Code

Helicone — Read: Quickstart — Helicone · Code

Opik (eval (Comet)) — Read: Quickstart — Opik · Code

OpenLLMetry (OTel for LLMs) — Read: What is OpenLLMetry? · Code

Evidently — Read: LLM Evaluation Quickstart — Evidently · Code

MLflow — Read: MLflow Tracking Quickstart · Code

LangWatch — Read: Quickstart — LangWatch · Code

Agenta (LLMOps platform) — Read: Quick Start — Agenta · Code

Weights & Biases (tracking + Weave) — Read: Weave Quickstart — W&B · Website

Serving & gateways

BentoML — Read: Hello world — BentoML · Code

OpenLLM — Read: OpenLLM in Action Part 1: The Basics · Code

Portkey AI Gateway (LLM gateway) — Read: AI Gateway — Portkey Docs · Code

GPTCache (semantic cache) — Read: GPTCache Tutorial · Code

Media generation (power-user)

ComfyUI — Read: Beginner's Guide to ComfyUI · Code

Stable Diffusion — Read: The Illustrated Stable Diffusion · Code

Whisper (speech-to-text) — Read: Getting Started with OpenAI Whisper · Code

text-generation-webui (Oobabooga UI) — Read: Exploring Oobabooga Text Generation Web UI · Code

FLUX (open image model) — Read: FLUX models — Black Forest Labs · Code

AUTOMATIC1111 (Stable Diffusion web UI) — Read: Features — SD web UI wiki · Code

Fooocus (one-click SDXL) — Read: Fooocus (overview) · Code

faster-whisper (fast speech-to-text) — Read: faster-whisper — README · Code

Kokoro TTS (open text-to-speech) — Read: Kokoro-82M — model card · Code

Chatterbox (TTS + voice cloning) — Read: Chatterbox — overview · Code

AudioCraft / MusicGen (music & audio gen) — Read: MusicGen — docs · Code

ACE-Step (music foundation model) — Read: ACE-Step — project page · Code

Wan (open video model) — Read: Wan-AI on Hugging Face · Code

LTX-Video (real-time video gen) — Read: LTX-Video — model card · Code

Embeddings & reranking

FlagEmbedding (BGE) (BGE embeddings + rerankers) — Read: BGE Documentation · Code

RAGatouille (ColBERT late-interaction) — Read: RAGatouille — README · Code

Level 4 — Research & scaleDistributed training, high-throughput inference engines, GPU kernels, quantization internals, RLHF at scale.Distributed / large-scale training

DeepSpeed — Read: Getting Started — DeepSpeed · Code

Megatron-LM — Read: Megatron-LM Framework Introduction · Code

Colossal-AI — Read: Quick Demo — Colossal-AI · Code

torchtitan (distributed training) — Read: torchtitan — Hugging Face · Code

nanotron (distributed training) — Read: Nanotron — Hugging Face · Code

GPT-NeoX — Read: A Comprehensive Guide to GPT-NeoX · Code

NVIDIA NeMo — Read: Overview — NVIDIA NeMo Framework · Code

OpenRLHF (RLHF at scale) — Read: Quick Start — OpenRLHF · Code

verl (RL post-training (PPO/GRPO)) — Read: verl Documentation · Code

High-throughput inference / serving

vLLM — Read: Quickstart — vLLM · Code

SGLang — Read: Quickstart — SGLang · Code

TensorRT-LLM — Read: Quick Start Guide — TensorRT-LLM · Code

Text Generation Inference (TGI) — Read: Hugging Face's Text Generation Inference Toolkit · Code

LMDeploy — Read: Quick Start — lmdeploy · Code

LightLLM — Read: Quick Start — Lightllm · Code

Triton Inference Server — Read: Quickstart — NVIDIA Triton Inference Server · Code

Text Embeddings Inference (embedding server) — Read: Quick Tour — Text Embeddings Inference · Code

SkyPilot (multi-cloud GPU) — Read: Quickstart — SkyPilot · Code

MLC LLM (cross-platform inference) — Read: MLC LLM — home · Code

KTransformers (CPU/GPU MoE offload) — Read: Introduction — KTransformers · Code

GPU kernels, quantization & efficiency

Triton (GPU kernels) — Read: Introducing Triton: Open-source GPU programming · Code

FlashAttention — Read: ELI5: FlashAttention · Code

llama.cpp — Read: How to Use llama.cpp to Run LLaMA Models Locally · Code

ExLlamaV3 (fast quantized inference) — Read: ExLlamaV2: The Fastest Library to Run LLMs · Code

MInference (long-context speedup) — Read: MInference: Million-Tokens Prompt Inference · Code

FlexGen (single-GPU inference) — Read: FlexGen: High-Throughput Generative Inference (paper) · Code

BitNet (1-bit LLMs) — Read: BitNet: Microsoft's 1-Bit LLMs on Your CPU · Code

NVIDIA Transformer Engine (FP8/FP4 training) — Read: Transformer Engine documentation · Code

Liger-Kernel (Triton training kernels) — Read: Liger Kernel (paper) · Code

FlashInfer (serving kernels) — Read: FlashInfer documentation · Code

torchao (PyTorch quant/sparsity) — Read: PyTorch Native Architecture Optimization · Code

GPTQModel (GPTQ/AWQ quantizer) — Read: GPTQ quantization (Transformers) · Code

Research evaluation / benchmarks

HELM (eval benchmark) — Read: Everything You Need to Know About HELM · Code

lm-evaluation-harness — Read: Evaluating LLM Accuracy with lm-evaluation-harness · Code

lighteval (eval) — Read: Quicktour — Lighteval · Code

SWE-bench (coding-agent benchmark) — Read: SWE-bench Leaderboards · Code

MTEB (embedding benchmark) — Read: MTEB: Massive Text Embedding Benchmark (paper) · Code