benchmark mcp servers

40 tagged benchmark, measured the same way as everything else here.

Browse within: agentic-browser 7browser-automation 7

codebase-memory-mcp

01

suyoumo/ClawProBench

MCP server Claude CodeCodexCursor +2

MCP server "codebase-memory-mcp" as configured in suyoumo/ClawProBench. Launched with codebase-memory-mcp.

823 7d ago A tokens not measured original Apache-2.0

unreal-mcp

02

ukanwat/aaabench

MCP server Claude CodeCodexCursor +2

MCP server "unreal-mcp", hosted remotely at 127.0.0.1, as configured in ukanwat/aaabench.

378 17d ago A tokens not measured original MIT

CodSpeed

03

CodSpeedHQ/codspeed

MCP server Claude CodeCodexCursor +2

MCP server "CodSpeed", hosted remotely at mcp.codspeed.io, as configured in CodSpeedHQ/codspeed.

280 3d ago A tokens not measured original Apache-2.0

spmind

04

tomtommyyuan/spmind

MCP server Claude CodeCodexCursor +2

MCP server "spmind" as configured in tomtommyyuan/spmind. Runs spmind.mcp_stdio with python.

277 1mo ago A tokens not measured original Apache-2.0

malicious

05

AIS2Lab/MCPSecBench

MCP server Claude CodeCodexCursor +2

MCP server "malicious" as configured in AIS2Lab/MCPSecBench. Launched with uv --directory /usr/games/mcp/MCPSecCode/mcpbench run maliciousadd.py.

35 6mo ago A tokens not measured original MIT

compute

06

AIS2Lab/MCPSecBench

MCP server Claude CodeCodexCursor +2

MCP server "compute" as configured in AIS2Lab/MCPSecBench. Launched with uv --directory /usr/games/mcp/MCPSecCode/mcpbench run addserver.py.

35 6mo ago A tokens not measured original MIT

cve

07

AIS2Lab/MCPSecBench

MCP server Claude CodeCodexCursor +2

MCP server "cve", hosted remotely at 127.0.0.1, as configured in AIS2Lab/MCPSecBench.

35 6mo ago A tokens not measured original MIT

folio

08

jfrog/agent-belt

MCP server Claude CodeCodexCursor +2

MCP server "folio", hosted remotely at 127.0.0.1, as configured in jfrog/agent-belt.

18 1mo ago A tokens not measured original Apache-2.0

intentforge

09

jzjzzzzzzz/IntentForge

MCP server Claude CodeCodexCursor +2

Intent-preserving deterministic CAD pipeline for editable, validated CadQuery models. Runs locally from the intentforge Python package.

6 21d ago A tokens not measured original Apache-2.0

openchainbench

10

ChainBench/OpenChainBench

MCP server Claude CodeCodexCursor +2

Live, neutral benchmarks for public RPC latency, oracles, bridges, perp DEX, and prediction markets. Remote server at openchainbench.com.

6 yesterday A tokens not measured original MIT

metrillm

11

MetriLLM/metrillm

MCP server Claude CodeCodexCursor +2

MCP server for MetriLLM — benchmark local LLMs from any AI coding assistant. Runs locally from the metrillm-mcp npm package.

5 3mo ago A tokens not measured original Apache-2.0

metrillm

12

MetriLLM/metrillm

MCP server Claude CodeCodexCursor +2

Benchmark local LLM models — speed, quality & hardware fitness verdict from any MCP client. Runs locally from the metrillm-mcp npm package.

5 3mo ago A tokens not measured original Apache-2.0

everything

13

PavelTkachenk0/ContextTax

MCP server Claude CodeCodexCursor +2

MCP server that exercises all the features of the MCP protocol. Runs locally from the @modelcontextprotocol/server-everything npm package.

5 2mo ago A tokens not measured original MIT

mcp-docs

14

0717376/cowork_bench

MCP server Claude CodeCodexCursor +2

The MCP documentation itself, served as an MCP server: lets the agent search the protocol spec and guides. Remote server at modelcontextprotocol.io.

5 23d ago A tokens not measured copy · 100% Apache-2.0

DanceNitra/agora

MCP server Claude CodeCodexCursor +2

MCP server "agora-memory-toolkit" as configured in DanceNitra/agora. Runs locally from the agora-memory-toolkit Python package.

3 today A tokens not measured original MIT

playwright

16

pleasedodisturb/web-agent-comparison

MCP server Claude CodeCodexCursor +2

MCP server "playwright" as configured in pleasedodisturb/web-agent-comparison. Launched with playwright-mcp.

2 19d ago A tokens not measured original MIT

browser-use

17

pleasedodisturb/web-agent-comparison

MCP server Claude CodeCodexCursor +2

MCP server "browser-use" as configured in pleasedodisturb/web-agent-comparison. Launched with browser-use --mcp.

2 19d ago A tokens not measured original MIT

lightpanda

18

pleasedodisturb/web-agent-comparison

MCP server Claude CodeCodexCursor +2

MCP server "lightpanda" as configured in pleasedodisturb/web-agent-comparison. Launched with lightpanda mcp.

2 19d ago A tokens not measured original MIT

codex

19

SSC-STUDIO/Ai-Model-Gateway

MCP server Claude CodeCodexCursor +2

MCP server "codex" as configured in SSC-STUDIO/Ai-Model-Gateway. Launched with codex mcp-server. Needs 1 environment variable to run.

2 7d ago A tokens not measured original MIT

llm-latency-tracker

20

mazamaka/llm-latency-tracker

MCP server Claude CodeCodexCursor

Measured latency, time to first token and uptime for 45 AI inference APIs, by region. Remote server at llmlatency.dev.

1 yesterday A tokens not measured

memtrust

21

RudrenduPaul/memtrust

MCP server Claude CodeCodexCursor

Independent, reproducible benchmark harness for agent-memory backends (MemPalace, Mem0, Zep/Graphiti, OpenViking): runs LongMemEval, LoCoMo, contradiction-detection, and a dozen other evals against all four and publishes the raw logs, not vendor-curated numbers. Runs locally from the memtrust-cli Python package.

1 6d ago A tokens not measured original Apache-2.0

distil

22

munhq/distil

MCP server Claude CodeCodexCursor

Context optimization middleware for LLM agents — a dynamic tool registry, result masking, token budgeting and smart compaction, so a long session stays inside its window instead of failing for length. Runs locally from the @munhq/distil npm package.

1 3d ago A tokens not measured

AiAgentKarl/llm-benchmark-mcp-server

MCP server Claude CodeCodexCursor

MCP Server for LLM comparison, benchmarks, and pricing — find the best model for any task. Runs locally from the llm-benchmark-mcp-server Python package.

1 5mo ago A tokens not measured original MIT

sudo-ai-git/mcp-benchmark-hygiene

MCP server Claude CodeCodexCursor

MCP server detecting pytest config-leakage that corrupts agent-benchmark grading (deterministic, no-LLM). Runs locally from the mcp-benchmark-hygiene Python package.

0 3d ago A tokens not measured original MIT