Go

1.1x faster.
Zero dependencies.

ZeroMCP vs mcp-go (mark3labs) — HTTP benchmarks.

Throughput
4.74K req/s ZeroMCP
vs.
4.20K req/s mark3labs
Memory
30 MB ZeroMCP
vs.
14 MB mark3labs

HTTP Performance — Head to Head

Same hello tool. Same methodology. 5-minute sustained load in Docker. Chi router for ZeroMCP, built-in HTTP for mcp-go.

req/s p50 p99 Memory CPU
ZeroMCP (Chi) 4.74K 0.18ms 0.45ms 30 MB 0.00%
mcp-go (mark3labs) 4.20K 0.17ms 0.67ms 14 MB 0.00%

The tradeoff

Choose mcp-go

If you want the leanest possible binary. 14 MB with built-in HTTP transport. No router dependency.

  • 14 MB binary — leanest option
  • Built-in HTTP transport
  • No router dependency required
Choose ZeroMCP

If you want to embed MCP into an existing Go service. Mount on Chi, Gin, or net/http — 1.1x faster with zero dependencies.

  • 0 dependencies — stdlib only
  • Chi + net/http + Gin
  • Built-in sandbox with enforced permissions
  • 4.74K req/s on Chi

Drop a .go file. It's an MCP tool.