Skip to main content

​
Python

​
Minimal install

pip install cascadeflow
Core dependencies: pydantic>=2.0.0, httpx>=0.25.0, tiktoken>=0.5.0, rich>=13.0.0.

​
With providers

pip install "cascadeflow[providers]"  # OpenAI + Anthropic + Groq
Individual providers: 
pip install "cascadeflow[openai]"      # OpenAI
pip install "cascadeflow[anthropic]"   # Anthropic
pip install "cascadeflow[groq]"        # Groq
pip install "cascadeflow[huggingface]" # Hugging Face
pip install "cascadeflow[together]"    # Together AI

​
With framework integrations

pip install "cascadeflow[langchain]"       # LangChain/LangGraph
pip install "cascadeflow[openai-agents]"   # OpenAI Agents SDK
pip install "cascadeflow[crewai]"          # CrewAI (Python 3.10+)
pip install "cascadeflow[google-adk]"      # Google ADK (Python 3.10+)

​
Local inference

pip install "cascadeflow[vllm]"  # vLLM (Python 3.10-3.13)
Ollama does not need a Python package — cascadeflow communicates with Ollama via HTTP at localhost:11434. Install Ollama separately from ollama.ai.

​
Everything

pip install "cascadeflow[all]"  # All providers + semantic routing

​
Development

git clone https://github.com/lemony-ai/cascadeflow.git
cd cascadeflow
pip install -e ".[dev]"

​
TypeScript

​
Core

npm install @cascadeflow/core

​
Framework packages

npm install @cascadeflow/langchain                # LangChain integration
npm install @cascadeflow/vercel-ai                 # Vercel AI SDK middleware
npm install @cascadeflow/n8n-nodes-cascadeflow     # n8n community node

​
Provider Setup

Set API keys as environment variables: 
export OPENAI_API_KEY="sk-..."
export ANTHROPIC_API_KEY="sk-ant-..."
export GROQ_API_KEY="gsk_..."
cascadeflow auto-detects available providers based on which API keys are set.

​
Verify Installation

python -c "import cascadeflow; print(cascadeflow.__version__)"

python -c "from cascadeflow import init, run, HarnessConfig, HarnessRunContext; print('OK')"

​
Next Step

Start observing your LLM calls with zero code changes. Quickstart: Observe Mode →