Flowise Integration
Use Antbase as the LLM backend in Flowise's visual chain builder. Drag and drop ChatOpenAI nodes with Antbase routing for RAG chatbots, agents, and document processing flows.
What this enables
- Visual LLM chain building with Antbase model routing
- RAG chatbots with document loaders and vector stores
- Agent workflows with tool calling and memory
- No-code setup through Flowise's drag-and-drop interface
Quick Start
1. Install Flowise
npx flowise startOr use Docker: docker run -d -p 3000:3000 flowiseai/flowise
2. Add a ChatOpenAI node
In the Flowise canvas, drag a ChatOpenAI node from the LLMs section.
3. Configure the node
| Field | Value |
|---|---|
| Connect Credential | Create new OpenAI API credential with your Antbase key |
| Model Name | ant:auto |
| BasePath | https://antbase.ai/v1 |
Ready to build
Connect the ChatOpenAI node to chains, agents, or memory nodes as usual.
Configuration
You can also set Antbase via environment variables when starting Flowise:
OPENAI_API_KEY=your-ant-api-key \
OPENAI_API_BASE=https://antbase.ai/v1 \
npx flowise startUsage Examples
RAG chatbot
Build a document Q&A chatbot:
- Add a PDF Loader and connect to a Recursive Text Splitter
- Connect the splitter to an In-Memory Vector Store with an embeddings node
- Add a Conversational Retrieval QA Chain
- Connect the ChatOpenAI node (with Antbase) and the vector store to the chain
Using the Flowise API
# After deploying a chatflow, call it via API:
curl -X POST http://localhost:3000/api/v1/prediction/<chatflow-id> \
-H "Content-Type: application/json" \
-d '{"question": "What does the document say about revenue?"}'Virtual Models
| Model | Best for | Notes |
|---|---|---|
| ant:auto | General use | Recommended for most chatflows |
| ant:fast | Simple Q&A | Fast responses for high-volume chatbots |
| ant:best | Complex analysis | Frontier models for demanding tasks |
| ant:free | Development | Zero cost while building flows |
| ant:reasoning | Multi-step logic | Chain-of-thought reasoning |
| ant:coder | Code generation | Code-optimized models |
See Virtual Models for the full list.
Troubleshooting
Model not found in dropdown
Type the model name manually in the Model Name field (e.g., ant:auto). The dropdown only lists OpenAI models but Flowise accepts any string.
BasePath not saving
In some Flowise versions, the BasePath field is in the "Additional Parameters" section of the ChatOpenAI node. Expand it to find the field.