Flowise

Build AI agents visually and route them through TokenMix.

Prepare TokenMix values

• API Key: your TokenMix key, usually starting with sk-tm-
• Chat model ID: copy it from TokenMix, for example
• Embedding model ID: needed only for RAG, for example
• Base Path: https://api.tokenmix.ai/v1

Start Flowise locally

npm install -g flowise
npx flowise start

Open http://localhost:3000 and create a Chatflow.

Configure the chat model

Use ChatOpenAI-Custom first because it lets you manually type a TokenMix model ID.

1. Add a ChatOpenAI-Custom node.
2. Create an OpenAI API Key credential and paste your TokenMix key.
3. Set Model Name to a TokenMix model ID.
4. Set Base Path to https://api.tokenmix.ai/v1.
5. Keep Streaming enabled.
6. Start with Temperature = 0.7.
7. Leave Base Options empty unless you need custom headers.

Configure embeddings for RAG

For vector search or RAG, add OpenAIEmbeddingsCustom or OpenAIEmbeddings:

• API Key: the same TokenMix key
• Model Name: a TokenMix embedding model ID
• Base Path: https://api.tokenmix.ai/v1

Do not use a chat model as an embedding model.

Test prompt

Explain TokenMix API in three short sentences and mention why it works well with Flowise.

Troubleshooting

• 401: wrong API key or extra spaces.
• 404/model not found: wrong Model Name.
• connection error: check Base Path.
• RAG fails: make sure the embedding node also uses TokenMix.
• model not in dropdown: use ChatOpenAI-Custom and type the model ID manually.