Matte blue claw gateway routing chat, document, and database workloads into code, inbox triage, and meeting-note outputs.

For cost-conscious OpenClaw, Hermes, and OpenCode users

Cost-efficient compute for agent workloads.

Point your OpenAI-compatible client at Blue Claw and run recurring agent work on lower-cost open models. Change one setting and keep your stack.

Get Started View Quick Start

Already have an account? Go to Console.

01 - The Problem

Recurring agent work gets expensive.

Agents run in loops. They review code, triage inboxes, summarize transcripts, and index documents, making the same kind of call thousands of times. At that volume the cost per call is what decides whether the workload is worth running, and on premium APIs it adds up fast.

02 - The Switch

Change the endpoint. Keep the workflow.

Blue Claw is an OpenAI-compatible endpoint. Point your client at it, keep your SDK and tool wiring, and route recurring work to lower-cost open models. It runs on a network of independent GPUs, so the same calls cost less. See how it works.

client = OpenAI(
    base_url="https://openai.blueclaw.network/v1",
    api_key="your-api-key",
)
  • Streaming in the standard SSE format
  • Tool calling with your existing wiring
  • Open-model presets for recurring agent work
  • Drop-in switch: change base_url, keep the stack

Run today

03 - Pricing

Predictable cost, no surprise bills.

Every new account starts with a $5 welcome credit. After that, you run on prepaid USD credits — buy what you use, no subscriptions. Per-request pricing is metered against a live price sheet, with per-key spend caps so agent workloads stay inside budget.

Workload Premium APIs Blue Claw
Recurring chat & agents Metered at frontier per-token rates Capable open models, lower marginal cost
Embeddings & indexing Per-token, adds up at scale Batch-friendly, lower cost
Image generation Premium per-image pricing Open image models
Heavy / async jobs Full price, no queueing discount Async job mode with cost ceilings in development

Live per-model pricing is on the Models page in the console. See the quick start to try it on one workload first.

04 - Trust & Privacy

Clear privacy boundaries, baked in.

Practical constraints are part of the product story, not a footnote.

See the privacy policy for details.

No content logging by default
Blue Claw does not log prompts, inputs, or outputs by default.
Operational metadata
Limited operational metadata is retained for abuse prevention, billing, reliability, and debugging.
Not every workload yet
No TEE, E2EE, or confidential-compute guarantee yet. Review sensitive or regulated workloads before using the service.