HermesAgent-MultiModel
A fully local, multi-model AI agent framework that routes tasks such as planning, coding, reasoning and vision to specialized models running under Ollama, then synthesizes their answers through a Mixture-of-Agents aggregator, tuned for consumer GPUs such as the RTX 4080 or 5080.
HermesAgent-MultiModel implements a local-first approach to multi-model inference: an eight-model roster covers tool use, fast responses, planning, reasoning, coding, embeddings and vision, and its Mixture-of-Agents orchestration queries several of those models in parallel and synthesizes their outputs through an aggregator rather than relying on a single model for every task. Task routing, personalities and tool-calling are handled through Hermes, configured against a local Ollama backend so no API keys are required and full 64K context windows stay local.
The repository is primarily configuration and documentation: a Hermes provider and routing configuration file, and an eleven-section optimization guide covering hardware choices, model selection and systemd tuning such as flash attention and KV cache quantization settings for Ollama.
It targets a GPU with at least 16 GB of VRAM, 48 GB or more of system RAM, and runs on Ubuntu, WSL2 or macOS.