How to Configure Ollama Local Models in Navin

Use Ollama when you want models to run on this machine with no cloud API key. Navin talks to Ollama through its local OpenAI-compatible endpoint (typically http://localhost:11434/v1).

What you will build

Ollama installed and running on your machine
at least one pulled model (for example llama3.2)
Navin pointed at Ollama under Settings → Providers
a model configuration pinned to the Ollama provider

Recommended path: Settings → Providers → Ollama

Open Settings → Providers → Ollama. The Local Ollama setup panel runs this flow:

Detect - checks whether Ollama is installed, whether the local server answers, which models are already pulled, and whether Navin already has a base URL configured.
Install - when Ollama is missing, offers a platform installer or a link to ollama.com/download. Install the Ollama desktop app from that site if the in-app installer is unavailable.
Start - starts the local server if it is installed but not answering.
Pull - downloads a recommended model (llama3.2, qwen2.5-coder:7b, mistral, gemma3:4b, nomic-embed-text) or any model id you type.
Configure / Use - writes the Ollama base URL when missing, creates a model configuration for the chosen model, and can make that configuration active. A custom base URL you already set (LAN host, non-default port) is preserved.

After Use, open Settings → Models if you want to review the active configuration, then send a chat message.

Recommended models

Model id	Role	Approx. size
`llama3.2`	General chat and light coding	~2 GB
`qwen2.5-coder:7b`	Local coding	~4.7 GB
`mistral`	Fast general-purpose	~4.1 GB
`gemma3:4b`	Small everyday tasks	~3.3 GB
`nomic-embed-text`	Free local embeddings / semantic search	~274 MB

Pull only what your GPU or RAM can hold. Start with llama3.2 if unsure.

Provider resolution notes

Pin the Ollama provider in the model configuration. Generic names such as llama3.2 do not always auto-route without a configured local base URL.
An empty Ollama block is not treated as configured. Use the Settings panel (or set the base URL) and pin the provider.
NVIDIA cloud nemotron models route to NVIDIA NIM when that key is present. Local Nemotron weights on Ollama use the Ollama provider.

Embeddings / semantic search

In Settings (or advanced tools config), point semantic search at Ollama with model nomic-embed-text after pulling that model from the Ollama panel.

Image generation

Ollama image models use a native API surface; chat still uses the /v1 base. When the chat base URL ends with /v1, image generation rewrites it automatically. See Image generation.

Troubleshooting

Symptom	Likely cause
Connection refused	Ollama not running, or wrong host/port in the base URL
Model not found	Model not pulled; use the panel Pull button
Very slow replies	Model too large for the machine; try a smaller tag
Install button unavailable	Download Ollama from ollama.com and reopen the panel
Navin still on a cloud model	Active configuration is not the Ollama one; open Settings → Models