Skip to content
modelread

Loading devices…

Remembered on this device. Multi-GPU: search “2x”, “8x”, “node”.

Support

Questions, bug reports and wrong estimates go to the public discussion board: huggingface.co/spaces/patrickbdevaney/modelread-mcp/discussions. Include the repo id, the device, the context length, and (for speed reports) what you measured and with which runtime.

Connection problems

The MCP endpoint is https://modelread.com/mcp (streamable HTTP, no key). If a model is new to the service, the first answer can be “queued” while its files are read from the Hub; ask again after about a minute.

Make the numbers better

Measured runs improve predictions for everyone: modelfit bench --submit (llama-bench) or modelfit harvest (llama-server logs).

Askwhat should I run?
Browsebest models per device
Devicesall hardware
Connect your agentMCP · API
How the numbers workcalibration