Ollama
Use models served by a local Ollama instance for embeddings and generation
The ollama provider calls an Ollama server through its OpenAI-compatible API.
What It Provides
| Component | Inputs | Notes |
|---|---|---|
| Embedder | Text | For example nomic-embed-text |
| Generator | Text, images | Images need a vision model; streaming and function tools where the model supports them |
Credentials
None. The endpoint defaults to http://127.0.0.1:11434; set url for another
host.
Example
{
"embedders": {
"ollama-text": { "provider": "ollama", "model": "nomic-embed-text" }
},
"generators": {
"ollama-chat": { "provider": "ollama", "model": "llama3.2" }
}
}
Ollama embeddings do not accept retrieval input types or query instructions. For a local model with those controls, use the Antfly provider.