inferbench
Benchmarks local LLM inference speed (tokens/sec) on your own hardware via MCP tools.
Install
PyPIstdio
uvx inferbench-cli Client config (Claude Desktop, Cursor, Windsurf and most MCP clients)
{
"mcpServers": {
"io-github-rudrendupaul-inferbench": {
"command": "uvx",
"args": [
"inferbench-cli"
]
}
}
} Source: official MCP Registry record · registry name io.github.RudrenduPaul/inferbench · published 2026-08-11 · updated 2026-08-11. Not an endorsement; review the code before connecting it to anything sensitive.