Ollama

Provider key: ollama
Support tier: runnable (local execution only)

Ollama enables running supported open-weight models on your own machine. It can reduce LenserFight-hosted data sharing and provider API cost for local development, but operators should still review Ollama's own model download, update, logging, and network behavior before using it for sensitive workflows.

How it works

LenserFight's local lenser connects to an Ollama instance running on your machine. The execution path is:

1. You start ollama serve locally (default port 11434).
2. LenserFight CLI detects the local endpoint and routes lens executions there instead of the cloud.
3. Results are returned to the platform and stored as normal.

Configuration

1. Install Ollama from ollama.com
2. Pull the model you want: ollama pull llama3.2
3. Start the server: ollama serve
4. In LenserFight, select Local as the execution target and pick your Ollama model

See the local models setup guide for full instructions.

Supported models

Any model pulled into your local Ollama installation is usable. Common choices:

Model	Key (when configured)	Notes
Llama 3.2	ollama/llama3.2	Strong general-purpose chat
Gemma 3	ollama/gemma3	Google's open model, good for reasoning
Mistral Nemo	ollama/mistral-nemo	Efficient instruction model
DeepSeek-R1	ollama/deepseek-r1	Reasoning-focused
Qwen 3	ollama/qwen3	Multilingual, strong tool use

Related

• Local models setup guide
• CLI getting started
• AI Providers overview