Ollama

Ollama is a supported execution provider on LenserFight for locally-hosted open-weight models. Rather than routing to a cloud API, LenserFight connects to a running Ollama instance through the gateway daemon. This enables private, offline-capable lenses and battles without sending data to an external provider.

Support tier

runnable — LenserFight has a direct runtime path or adapter for this provider (via the local gateway daemon).

Upstream docs

ollama.com

Models on LenserFight

No models are registered in the default catalog yet. Configure via BYOK.

Pull any model with ollama pull <model> locally, then reference it by its Ollama tag in the lens model_id field (e.g. ollama/llama3.3, ollama/mistral). The gateway adapter proxies the call to your local Ollama instance.

Usage notes

• Start the LenserFight gateway daemon (lenserfight gateway start) before running lenses backed by Ollama; the daemon manages the local connection.
• Ollama models are subject to local hardware constraints — context window and throughput depend on available VRAM and the model quantisation level.