mcpcost

io.github.pipeworx-io/ai-model-experiments

Ai Model Experiments

AI Model Experiments MCP ('Model Lab') — run the same prompts across many

alive probed 2026-10-15 · HTTP 200

49tools
40ktokens in Claude Opus 5.5
41ktokens in Claude Code
19.9 %of a 200k context window
24ko200k_base estimate
165 msto initialize
139 msto list tools
Endpoint
https://gateway.pipeworx.io/ai-model-experiments/mcp (streamable-http)
Host
gateway.pipeworx.io
Protocol
2025-03-26
Reports itself as
pipeworx-gateway 0.1.0
Capabilities
prompts, resources, tools
Instructions
732 characters
Claude count
Claude Opus 5.5, measured 2026-10-15. Counted again when the tool definitions change. Method.
Repository
https://github.com/pipeworx-io/mcp-ai-model-experiments
Website
https://pipeworx.io/packs/ai-model-experiments
Registry
active, updated 2026-09-25 · registry entry
Reachable
16 of the last 16 probes

Most expensive tools

ToolTokens (o200k)Share of this server
polymarket_kalshi_spread1 7387 %
bet_research1 2255 %
polymarket_edges1 1755 %
subscribe1 1345 %
company_facts1 0935 %

History

We add a row when the result changes. A server with no change keeps its last row.

ProbedResultInitToolsClaudeo200k
2026-10-15 10:58alive76 ms4940k24k
2026-10-14 19:12alive170 ms4940k24k
2026-10-14 09:24alive131 ms4940k24k
2026-10-13 11:17alive115 ms4838k23k
2026-10-12 11:47alive158 ms4838k23k
2026-10-11 10:27alive1.2 s4838k23k
2026-10-10 10:19alive120 ms4838k23k
2026-10-09 11:21alive90 ms4837k23k
2026-10-08 11:16alive90 ms4434k20k
2026-10-07 11:13alive322 ms4434k20k
2026-10-05 15:26alive165 ms4434k20k
2026-10-05 00:44alive57 ms4420k
2026-10-04 10:06alive78 ms4420k
2026-10-04 08:00alive54 ms4420k
2026-10-03 20:39alive95 ms4420k
2026-10-03 17:54alive139 ms4420k

Badge and feed

mcpcost: alive

[![mcpcost](https://mcpcost.ulehla.dev/badge/io.github.pipeworx-io/ai-model-experiments.svg)](https://mcpcost.ulehla.dev/s/io.github.pipeworx-io/ai-model-experiments)

An entry for every probe that changed the result, the tool count or the token cost: Atom feed. Add it to a feed reader or a Slack RSS app to hear when the server stops answering or its tool set grows.