TDSPRO developer docs

Overview

TDSPRO is a single, OpenAI-compatible gateway. You get one API key and one base URL that route every request to the right model — with smart routing, automatic fallback, a token-saving cache, per-key budgets, and full usage analytics. Because the API matches the OpenAI spec exactly, any existing client, SDK, or tool works without code changes.

Quickstart

Three steps and you are live. The snippets below are complete and runnable — paste your key and go.

1. 1Create a free API key in your dashboard. It starts with sk-hub-.
2. 2Set your client base URL to https://api.tdspro.lol/v1.
3. 3Call any model by alias (smart, fast, code…) or by full name (gpt-4.1).

Pick your language. All three do the same thing.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.tdspro.lol/v1",
  apiKey: process.env.AIHUB_API_KEY, // sk-hub-...
});

const res = await client.chat.completions.create({
  model: "smart",                    // alias → best model for the job
  messages: [{ role: "user", content: "Hello from TDSPRO!" }],
});

console.log(res.choices[0]?.message.content);

from openai import OpenAI
import os

client = OpenAI(
    base_url="https://api.tdspro.lol/v1",
    api_key=os.environ["AIHUB_API_KEY"],  # sk-hub-...
)

res = client.chat.completions.create(
    model="smart",                        # alias → best model for the job
    messages=[{"role": "user", "content": "Hello from TDSPRO!"}],
)
print(res.choices[0].message.content)

curl https://api.tdspro.lol/v1/chat/completions \
  -H "Authorization: Bearer $AIHUB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "smart",
    "messages": [{"role": "user", "content": "Hello from TDSPRO!"}]
  }'

Authentication

Authenticate every request with a Bearer token — your TDSPRO key, which starts with sk-hub-. The OpenAI SDKs set this header for you when you pass apiKey.

Authorization: Bearer sk-hub-...

# Keep keys in environment variables — never in source control.
export AIHUB_API_KEY="sk-hub-..."      # macOS / Linux
$env:AIHUB_API_KEY = "sk-hub-..."      # Windows PowerShell

Key security

• ✓Store keys in environment variables or a secret manager — never in source control or client-side code.
• ✓Create a separate key per project or environment, and set a spend cap and rate limit on each from the dashboard.
• ✓Rotate or revoke a key instantly in the dashboard if it leaks. Old keys stop working immediately.
• ✓Every request runs over HTTPS (TLS), so your key and payloads are encrypted in transit.

Models & aliases

Call a friendly alias and let smart routing choose the best available model, or pass an exact model name. List everything your key can reach from the /models endpoint.

Capability aliases

These route automatically to the best model for the task — and fall back if one is busy.

Named models

Prefer an exact flagship? Call it by name. Flagship-class models — including GPT-4.1 and GPT-4o-class — are available directly.

List available models

Returns an OpenAI-style list. Each id can be passed straight into the "model" field.

curl https://api.tdspro.lol/v1/models \
  -H "Authorization: Bearer sk-hub-..."

{
  "object": "list",
  "data": [
    { "id": "smart",  "object": "model", "owned_by": "ai-hub" },
    { "id": "fast",   "object": "model", "owned_by": "ai-hub" },
    { "id": "code",   "object": "model", "owned_by": "ai-hub" },
    { "id": "vision", "object": "model", "owned_by": "ai-hub" }
    // ...and every named model your key can reach
  ]
}

How to verify the model is real

Endpoints

The core OpenAI surface is supported. All paths are relative to the base URL and require the Authorization header.

Chat completions

Request

curl https://api.tdspro.lol/v1/chat/completions \
  -H "Authorization: Bearer sk-hub-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "smart",
    "messages": [
      { "role": "system", "content": "You are concise." },
      { "role": "user",   "content": "Explain HTTP in one sentence." }
    ],
    "temperature": 0.7,
    "max_tokens": 256
  }'

Response

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "model": "gpt-4o",                 // the real model that answered
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "..." },
      "finish_reason": "stop"
    }
  ],
  "usage": { "prompt_tokens": 23, "completion_tokens": 14, "total_tokens": 37 }
}

The "model" field in the response is the real engine that answered — useful for logging and cost attribution.

Embeddings

curl https://api.tdspro.lol/v1/embeddings \
  -H "Authorization: Bearer sk-hub-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-embedding-001",
    "input": "The quick brown fox"
  }'

Returns an OpenAI-style data array of float vectors. Use them for semantic search and RAG.

Streaming

Set "stream": true to receive tokens as they are generated over Server-Sent Events (SSE), exactly like the OpenAI API. The stream ends with a [DONE] sentinel. The OpenAI SDKs handle this for you — just iterate the response.

With the SDK

from openai import OpenAI

client = OpenAI(base_url="https://api.tdspro.lol/v1", api_key="sk-hub-...")

stream = client.chat.completions.create(
    model="smart",
    messages=[{"role": "user", "content": "Write a haiku about caching."}],
    stream=True,
)
for chunk in stream:
    delta = chunk.choices[0].delta.content or ""
    print(delta, end="", flush=True)

Raw SSE wire format

data: {"choices":[{"delta":{"content":"Hello"}}]}

data: {"choices":[{"delta":{"content":" world"}}]}

data: {"choices":[{"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Each event is a data: line with a JSON delta. Concatenate the delta.content fields to rebuild the message.

Errors & status codes

Errors use standard HTTP status codes and a consistent JSON shape, so OpenAI-style error handling works unchanged.

Error shape

{
  "error": {
    "type": "rate_limit_exceeded",
    "message": "You hit your per-key rate limit. Retry shortly.",
    "code": 429
  }
}

Status codes

Retry on 429 & 5xx

On a 429 or transient 5xx, retry with exponential backoff. Automatic fallback often resolves it before you even see the error.

async function withRetry(fn, tries = 4) {
  for (let i = 0; i < tries; i++) {
    try {
      return await fn();
    } catch (e) {
      // Retry only on 429 and transient 5xx.
      const status = e?.status ?? e?.response?.status;
      if (![429, 500, 502, 503].includes(status) || i === tries - 1) throw e;
      await new Promise((r) => setTimeout(r, 2 ** i * 500)); // 0.5s, 1s, 2s, 4s
    }
  }
}

Rate limits & quotas

Rate limits and budgets are set per key in your dashboard, not hard-coded globally — so you control them. Limits scale with your plan.

Caching & token savings

Identical requests are served from a token-saving cache, so repeat traffic costs less and returns instantly. It is the same model and the same answer — you simply pay for fewer of them.

Caching keys on the exact request (model + messages + parameters). Change any field — temperature, a message — and you get a fresh generation.

Examples

Copy-paste starters for the most common tools. Replace sk-hub-... with your key.

from openai import OpenAI
import os

client = OpenAI(
    base_url="https://api.tdspro.lol/v1",
    api_key=os.environ["AIHUB_API_KEY"],  # sk-hub-...
)

res = client.chat.completions.create(
    model="smart",                        # alias → best model for the job
    messages=[{"role": "user", "content": "Hello from TDSPRO!"}],
)
print(res.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.tdspro.lol/v1",
  apiKey: process.env.AIHUB_API_KEY,
});

const r = await client.chat.completions.create({
  model: "code",                       // tuned for programming
  messages: [{ role: "user", content: "Write a debounce() in TypeScript." }],
});
console.log(r.choices[0].message.content);

curl https://api.tdspro.lol/v1/chat/completions \
  -H "Authorization: Bearer $AIHUB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "smart",
    "messages": [{"role": "user", "content": "Hello from TDSPRO!"}]
  }'

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "aihub": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "TDSPRO",
      "options": {
        "baseURL": "https://api.tdspro.lol/v1",
        "apiKey": "sk-hub-..."
      },
      "models": {
        "smart": { "name": "TDSPRO — Flagship" },
        "fast":  { "name": "TDSPRO — Fast" },
        "code":  { "name": "TDSPRO — Code" }
      }
    }
  },
  "model": "aihub/smart"
}

Settings → Models → OpenAI API Key
  Override OpenAI Base URL:  https://api.tdspro.lol/v1
  OpenAI API Key:            sk-hub-...
  Custom model name:         smart   (also add: fast, code, vision)

FAQ

Troubleshooting

Get help

Stuck or have a question? We respond fast — usually within a few hours.

Or reach us directly — Telegram is fastest: Telegram @tdspro_admin · [email protected]