Agent Models

Opensperm supports a wide variety of open-weight and proprietary models that can be run locally within your Agent Pod.

Supported Architectures

You can hot-swap models depending on the task requirements. We provide optimized inference engines (like vLLM and llama.cpp) pre-installed in the runtime.

Large Language Models (LLMs)

Llama 3 (8B, 70B), Mistral, Mixtral, Qwen, and custom fine-tunes.

Vision Models

LLaVA, Qwen-VL, and Stable Diffusion for image generation.