PROXYSTACKS RELAY

Ship AI featureswithout provider lock-in.

Use one OpenAI-compatible endpoint to route requests across models, track spend, and keep fallbacks ready when providers slow down.

Start free View docs

OpenAI-compatible

Usage and cost logs

Fallback routing

relay.dispatch

curl https://proxystacks.com/v1/chat/completions \
  -H 'Authorization: Bearer sk-relay-••••••••' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "deepseek-chat",
    "messages": [
      { "role": "user", "content": "Hello, Relay." }
    ],
    "stream": true
  }'

200 OK

model

relay-mini

response

Hello! Relay is ready
to route your request.

tokens

latency

624ms

50.8K

Requests routed

Active models

756ms

Avg latency

32.4%

Cost saved

OpenAI

online

DeepSeek

online

Gemini

standby

One Key To Every Model

◎OpenAI

◇DeepSeek

✦Gemini

◈Qwen

Core Capabilities

Everything your AI model traffic needs, managed through Relay.

From provider access and routing to billing and logs, ProxyStacks Relay turns model infrastructure into an operable product surface.

OpenAI-compatible API

Route OpenAI, DeepSeek, Gemini, and local providers through one stable endpoint on proxystacks.com.

Policy-based routing

Choose providers by model, cost, availability, and environment without changing application code.

Transparent cost control

Track pricing rules, balances, and request-level spend in a ledger your team can audit.

Usage intelligence

Capture tokens, latency, status codes, cost, and request IDs across every model call.

Project-scoped keys

Issue separate keys for products, teams, and environments with clear operational boundaries.

Secure by default

Hash keys, limit usage, and keep production traffic protected with sensible gateway defaults.

Workflow

Go from provider setup to cost optimization in four steps.

Keep the developer API simple while Relay handles the operational control plane for your team.

Connect

Add model providers, pricing rules, model aliases, and environment-level access.

Route

Point your app to ProxyStacks Relay and let the gateway handle auth, mapping, and forwarding.

Observe

Review live requests, tokens, spend, latency, and errors from one operational console.

Optimize

Tune model strategy, rate limits, and budgets with real usage data instead of guesswork.

Agent products

Unify tool calls, streaming responses, and multi-model fallback behind one API surface.

Internal AI platforms

Share model access across teams while separating keys, quotas, logs, and budgets.

Development sandboxes

Run complete request flows with a mock provider before connecting production vendors.

Ready to Relay

Switch models without rebuilding your product.

Create your first Relay API key on proxystacks.com and decouple your application from single-provider risk.

Create a free key

Streaming first

Keep chat, agents, and tool calls moving through standard SSE responses.

Spend clarity

Trace balances, pricing rules, request logs, and cost changes from one place.

Operational guardrails

Store hashed keys and tighten limits progressively as traffic grows.