One gateway · multiple model families

One API gateway for your AI models.

Use compatible AI APIs through api.zrelay.net. Create Zrelay API keys, set spending limits and manage your active Base credit in one account.

3model families
%up to, on Scale
1 accountBase service credit
gateway · metered
# point any client at api.zrelay.net $ export ANTHROPIC_BASE_URL="https://api.zrelay.net" $ curl https://api.zrelay.net/v1/messages \ -H "x-api-key: zr_live_..." \ -H "anthropic-version: 2023-06-01" \ -H "content-type: application/json" \ -d '{ "model": "claude-opus-5", "max_tokens": 256, "messages": [{"role":"user","content":"Hello"}] }'
200 OK ·  Claude-format response · claude-opus-5
G $19 → $75 of Zrelay service credit
C 87% below Zrelay service-credit face value
Built for developers

Your keys. Your spending controls.

  • Create and revoke Zrelay-issued keys in your account portal.
  • Set a spending quota for each paid key.
  • Share prepaid credit across model families within each credit class.
  • See metered usage, token counts and pending reservations.
  • Start with a free trial before buying API usage.
  • One-time BTC purchases through BTCPay, subject to current availability; no subscription.
  • Keep compatible request formats and change your base URL.

Requests pass through Zrelay to upstream inference suppliers. Read about data processing and refund coverage.

Supported models

Your models. One place.

Use Claude, GPT, Grok, DeepSeek, Gemini, GLM and Kimi on Base. Each request uses the same Base service-credit balance.

The Base collection

Claude, GPT, Grok, DeepSeek, Gemini, GLM and Kimi keys share your Base balance.

AnthropicClaude
claude-opus-5 · claude-opus-5-5 · claude-sonnet-5

Opus and Sonnet through the Anthropic Messages format, with Claude Code support.

Claude docs
OpenAIGPT
gpt-6-astra · gpt-6-sol · gpt-5.6-terra · gpt-5.5 · GPT Image 2.5

Astra, Terra, Sol, Luna and GPT-5.5 through Responses and Chat APIs, plus Flare and Sunburst for image generation. Luna is temporarily unavailable.

GPT & image docs
xAIGrok
grok-4.6 · grok-4.5-latest

Grok 4.6 and Grok 4.5 through the OpenAI-compatible Chat Completions endpoint, with Grok Build support.

Grok docs
Choose your credit tier once. Manage keys and providers from your account. Choose a tier
A checkable example

See how the 33% is calculated.

Base uses one customer balance and published per-model rates. Primary execution uses a 1.0x customer multiplier; an eligible fallback uses 1.5x while native provider accounting stays separate.

Compare enabled model rates
Primary customer multiplier1.0×
Eligible fallback customer multiplier1.5×

Native provider units remain separate from customer service credit.

Why it costs less
Available inference capacity, managed through Zrelay.

Zrelay sources available inference capacity through independent upstream suppliers and capacity partners. Our primary supplier describes this as including unused allocation from startups and developers. Zrelay brings that capacity behind its own gateway, authentication, usage metering and billing. You receive Zrelay-issued credentials; upstream credentials are never distributed to customers. See how requests and prepaid usage work.

Early tester feedback

Notes from private testers.

These comments come from people invited to test Zrelay. They are early feedback, not paying-customer reviews or purchase ratings.

Contact

Questions before buying?
Ask directly.

For setup, billing or service questions, email Zrelay support. Support details and what to include.

Email support@zrelay.net

Email Zrelay support

  • Pre-sale questions: models, context limits, what a tier really buys.
  • Something not working? Include your account ID, the key label, and the exact error line. Never send an API key, account login code or receipt link.
  • Setup help for Claude Code, Codex, and Grok Build.