Skip to Content
API ReferenceProxy API

Proxy API Reference

Technical documentation for the AiSpendTrack proxy API.

Base URL

https://proxy.aispendtrack.com

Authentication

All requests must include your AiSpendTrack API key in headers:

x-aispendtrack-key: ask_your_key_here

Endpoints

OpenAI Proxy

Base URL:

https://proxy.aispendtrack.com/v1

Supported endpoints:

  • /v1/chat/completions - Chat completions
  • /v1/completions - Text completions
  • /v1/embeddings - Embeddings
  • /v1/images/generations - Image generation

Example request:

curl https://proxy.aispendtrack.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_OPENAI_KEY" \
  -H "x-aispendtrack-key: YOUR_AISPENDTRACK_KEY" \
  -d '{
    "model": "gpt-4",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Anthropic Proxy

Base URL:

https://proxy.aispendtrack.com/anthropic/v1

Supported endpoints:

  • /anthropic/v1/messages - Messages API
  • /anthropic/v1/complete - Completions (legacy)

Example request:

curl https://proxy.aispendtrack.com/anthropic/v1/messages \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_ANTHROPIC_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "x-aispendtrack-key: YOUR_AISPENDTRACK_KEY" \
  -d '{
    "model": "claude-3-opus-20240229",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Optional Headers

Customer Tracking

Track costs per customer:

x-customer-id: customer_123

The customer ID can be any string (user ID, email, account number).

Feature Tracking

Track costs per feature:

x-feature: chat-support

The feature name can be any string identifying the feature.

Combined Example

curl https://proxy.aispendtrack.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_OPENAI_KEY" \
  -H "x-aispendtrack-key: YOUR_AISPENDTRACK_KEY" \
  -H "x-customer-id: user_789" \
  -H "x-feature: content-generation" \
  -d '{
    "model": "gpt-4",
    "messages": [{"role": "user", "content": "Write a blog post"}]
  }'

This call will be tracked with:

  • Model: gpt-4
  • Customer: user_789
  • Feature: content-generation

Response Headers

AiSpendTrack adds these headers to responses:

x-aispendtrack-cost: 0.0045
x-aispendtrack-tokens: 225
x-aispendtrack-latency: 1234

Headers:

  • x-aispendtrack-cost - Cost in USD
  • x-aispendtrack-tokens - Total tokens used
  • x-aispendtrack-latency - Proxy latency in milliseconds

These headers are informational only. Full details appear in your dashboard.

Error Responses

401 Unauthorized

Cause: Missing or invalid AiSpendTrack API key

Response:

{
  "error": {
    "message": "Invalid or missing x-aispendtrack-key header",
    "type": "authentication_error",
    "code": "invalid_api_key"
  }
}

Solution: Verify your API key is correct and in headers.

429 Rate Limit Exceeded

Cause: You’ve exceeded your rate limit or monthly limit

Response:

{
  "error": {
    "message": "Rate limit exceeded",
    "type": "rate_limit_error",
    "code": "rate_limit_exceeded",
    "retry_after": 3600,
    "current_usage": 10000,
    "limit": 10000,
    "upgrade_url": "https://app.aispendtrack.com/upgrade"
  }
}

Headers:

retry-after: 3600
x-ratelimit-limit: 10000
x-ratelimit-remaining: 0
x-ratelimit-reset: 1708272000

Solution:

  • Wait for retry-after seconds
  • Or upgrade your plan

502 Bad Gateway

Cause: OpenAI/Anthropic is down or unreachable

Response:

{
  "error": {
    "message": "Upstream provider temporarily unavailable",
    "type": "service_error",
    "code": "upstream_error"
  }
}

Solution: Retry with exponential backoff.

Rate Limits

TierRate LimitMonthly Limit
Free1,000 req/hour10,000 calls
Pro10,000 req/hourUnlimited
EnterpriseCustomUnlimited

Rate limit headers:

x-ratelimit-limit: 1000
x-ratelimit-remaining: 847
x-ratelimit-reset: 1708272000

Streaming

Streaming is fully supported and works identically to direct API calls.

OpenAI streaming:

const stream = await openai.chat.completions.create({
  model: "gpt-4",
  messages: [{ role: "user", content: "Write a story" }],
  stream: true
});
 
for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}

Anthropic streaming:

with client.messages.stream(
    model="claude-3-opus-20240229",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Write a story"}]
) as stream:
    for text in stream.text_stream:
        print(text, end='', flush=True)

Streaming responses are tracked the same as non-streaming. Costs appear after stream completes.

Latency

Average latency overhead: <300ms

Latency breakdown:

  • Header processing: ~5ms
  • Database logging (async): ~50ms (doesn’t block response)
  • Network overhead: ~50-100ms
  • Total added latency: ~60ms on average

The proxy forwards your request immediately and logs metadata asynchronously.

Data Logged

For each API call, we log:

{
  "id": "call_abc123",
  "timestamp": "2024-02-18T14:30:00Z",
  "customer_id": "cust_123",
  "provider": "openai",
  "model": "gpt-4",
  "endpoint": "/v1/chat/completions",
  "prompt_tokens": 150,
  "completion_tokens": 75,
  "total_tokens": 225,
  "cost_usd": 0.0045,
  "latency_ms": 1234,
  "status_code": 200,
  "error_type": null,
  "customer_id": "user_789",
  "feature": "content-generation"
}

We never log:

  • ❌ Your prompts
  • ❌ AI responses
  • ❌ API keys (yours or OpenAI/Anthropic)
  • ❌ User data in prompts

Health Check

Endpoint:

GET /health

Response:

{
  "status": "ok",
  "uptime": 1234567,
  "version": "1.0.0"
}

Use for monitoring or load balancer health checks.

Regional Endpoints

Currently, we have one global endpoint:

https://proxy.aispendtrack.com

Coming soon:

  • 🔜 EU endpoint (GDPR compliance)
  • 🔜 Asia-Pacific endpoint (lower latency)

Enterprise customers can request dedicated endpoints.

SDK Support

Works with official SDKs:

Supported:

  • ✅ openai (Node.js, Python)
  • ✅ @anthropic-ai/sdk (Node.js)
  • ✅ anthropic (Python)
  • ✅ Any HTTP client (Go, Ruby, Java, etc.)

Configuration:

  • Change baseURL / base_url
  • Add x-aispendtrack-key header

No custom SDK required!

Next Steps

Last updated on
AnalyticsDashboard API