# DumboDhruvi/ModelFit-MCP

Source: https://glimind.com/server/github%2Fdumbodhruvi%2Fmodelfit-mcp
Registry id: `github/dumbodhruvi/modelfit-mcp`
Status: unknown

## Tools (9)

- `benchmark_models`: Run live accuracy and latency benchmarking on candidate models.
- `ensemble_predict`: Execute ensemble prediction combining multiple models.
- `find_ensemble_models`: Find ultra-fast models suitable for low-latency ensembling.
- `get_active_model_status`: Inspect currently loaded model and target device.
- `get_hardware_specs`: Detect host CPU, RAM, and GPU/VRAM headroom.
- `get_integration_code`: Get drop-in Python inference code.
- `recommend_and_scaffold`: 1-shot model search, hardware check, and code scaffolding.
- `search_compatible_models`: Search Hugging Face models strictly filtered by hardware fit.
- `swap_active_model`: Hot-swap the active model with VRAM-safe memory purging.

Measured by Glimind (https://glimind.com): MCP handshake + tool list only, no tool is ever called. CC BY 4.0.
