ZeroGPU Router

Cut your OpenClaw agent's inference costs. Route trivial AI tasks — summarize, classify, redact PII, extract JSON, generate follow-ups, short chat — to small/nano models instead of burning frontier-model tokens.

What it does

Your OpenClaw agent keeps doing the heavy reasoning. Routine tasks get offloaded to ZeroGPU's small models — typically 100–1000× cheaper per call.

• 11 task-specific MCP tools (zerogpu_summarize, zerogpu_classify_iab, zerogpu_redact_pii, …)
• A routing skill that teaches your agent when to call each tool
• Per-call savings logged with model, latency, and a real savings_usd figure
• Hosted MCP server at https://mcp.zerogpu.ai/mcp — no infra to run

Install

Run this in your OpenClaw terminal:

tmpdir=$(mktemp -d) && cd "$tmpdir" \
  && npm pack [email protected] \
  && tar -xzf zerogpu-openclaw-plugin-*.tgz \
  && cd package \
  && openclaw plugins install ./

Set up (2 steps)

1. Get your API key

Sign in at platform.zerogpu.ai and create a project to grab an API key + project ID.

2. Register the hosted MCP server

In your OpenClaw shell:

openclaw mcp set zerogpu '{
  "url": "https://mcp.zerogpu.ai/mcp",
  "transport": "streamable-http",
  "headers": {
    "x-api-key": "<your-api-key>",
    "x-project-id": "<your-project-id>"
  }
}'

Verify:

openclaw mcp show zerogpu --json

That's it. Your agent will now route trivial tasks through ZeroGPU automatically.

Try it

Ask your agent:

summarize this paragraph: Renewable energy adoption is accelerating globally, driven by falling solar and wind costs.

The agent calls zerogpu_summarize (running on t5-small) instead of the host model and replies with the summary plus a savings line.

The 11 tools you get

| Tool | Workload | Backing model | |---|---|---| | zerogpu_classify_iab | IAB topic classification | zlm-v1-iab-classify-edge | | zerogpu_summarize | TL;DRs, abstracts, meeting summaries | t5-small | | zerogpu_classify_zero_shot | Classify against a flat label list | deberta-v3-small | | zerogpu_extract_entities | People, places, companies, dates, custom entities | gliner2-base-v1 | | zerogpu_extract_json | Pull structured fields into grouped JSON | gliner2-base-v1 | | zerogpu_classify_structured | Multi-axis schema classification | gliner2-base-v1 | | zerogpu_redact_pii | Mask emails, phones, names, addresses, other PII | gliner-multi-pii-v1 | | zerogpu_extract_pii | Extract PII grouped by category | gliner-multi-pii-v1 | | zerogpu_generate_followups | Generate follow-up questions from a passage | zlm-v1-followup-questions-edge | | zerogpu_chat | Short small-model chat replies | LFM2.5-1.2B-Instruct / -Thinking | | zerogpu_health | Verify the ZeroGPU backend | — |

Every tool returns { <task fields>, model, usage, savings }.

Watch your savings

Live dashboard at platform.zerogpu.ai — token usage, latency, per-tool savings, broken down by agent and time range.

Links

• Website: https://zerogpu.ai
• Dashboard: https://platform.zerogpu.ai
• Source: https://github.com/zerogpu/zerogpu-router
• Full setup guide: https://github.com/zerogpu/zerogpu-router/tree/main/agents/openclaw
• Issues / contact: [email protected]

License

MIT — see LICENSE.