OpenAI-Compatible API

⚠️ BETA: GMTech's API is currently in beta testing. We welcome your feedback!

GMTech provides a fully OpenAI-compatible API as a drop-in replacement for OpenAI's API, with access to 15+ AI providers.

Why GMTech?

• Single API, Multiple Providers - Access OpenAI, Anthropic, Google, Meta, and more
• No Vendor Lock-in - Switch models without changing code
• Transparent Pricing - Real-time cost tracking; provider cost + a small 2% API service fee, shown in every response
• OpenAI Compatible - Use existing OpenAI SDK code unchanged

Quick Start

1. Install OpenAI SDK

pip install openai

2. Replace Base URL

from openai import OpenAI

# Instead of OpenAI...
# client = OpenAI(api_key="sk-...")

# Use GMTech
client = OpenAI(
    base_url="https://app.gmtech.com/v1",
    api_key="gmtech_your_api_key"
)

# Everything else works the same!

Chat Completions

Basic Chat

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is Python?"}
    ]
)

print(response.choices[0].message.content)

Parameters

Cost in Response

Every chat completion and image generation response includes a gmtech.cost_usd field with the exact USD amount charged for that request:

{
  "id": "chatcmpl-123",
  "choices": [...],
  "usage": { "prompt_tokens": 120, "completion_tokens": 80, "total_tokens": 200 },
  "gmtech": { "cost_usd": 0.000312 }
}

For image generation:

{
  "created": 1677652288,
  "data": [{ "url": "https://..." }],
  "gmtech": { "cost_usd": 0.04 }
}

Use Any Model

# OpenAI
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}]
)

# Anthropic Claude
response = client.chat.completions.create(
    model="claude-sonnet-46",
    messages=[{"role": "user", "content": "Hello!"}]
)

# Google Gemini
response = client.chat.completions.create(
    model="google-gemini-25-pro",
    messages=[{"role": "user", "content": "Hello!"}]
)

Image Generation

response = client.images.generate(
    model="gpt-image-2",  # DALL-E 3
    prompt="A futuristic city with flying cars",
    size="1024x1024"
)

print(response.data[0].url)

Embeddings

Generate vector embeddings for text using the OpenAI SDK or a direct HTTP call. Embeddings are useful for semantic search, RAG retrieval, and deduplication.

response = client.embeddings.create(
    model="text-embedding-3-small",
    input="The Methow Valley trail crew meets every Saturday"
)

vector = response.data[0].embedding  # list of 1536 floats

Batch multiple texts in one call:

response = client.embeddings.create(
    model="text-embedding-3-small",
    input=["first document", "second document", "third document"]
)

for item in response.data:
    print(item.index, len(item.embedding))  # 0 1536, 1 1536, 2 1536

Parameters

Available Models

Response

{
  "object": "list",
  "data": [
    {"object": "embedding", "index": 0, "embedding": [0.021, -0.113, 0.049, ...]}
  ],
  "model": "text-embedding-3-small",
  "usage": {"prompt_tokens": 8, "total_tokens": 8},
  "gmtech": {"cost_usd": 0.00000016}
}

Sentiment Analysis

Analyze the emotional tone of any text block. Returns 10 sentiment dimensions scored 0–100, powered by gpt-4o-mini structured outputs.

import requests, os

response = requests.post(
    "https://app.gmtech.com/v1/sentiment",
    headers={"Authorization": f"Bearer {os.environ['GMTECH_API_KEY']}"},
    json={"text": "I'm really excited about this new product launch!"}
)

print(response.json()["sentiment"])
# {'Positive': 82, 'Negative': 5, 'Neutral': 13, 'Joyful': 74, 'Angry': 2, ...}

Parameters

Response

Returns a sentiment object with 10 dimensions — each an integer 0–100:

Positive, Negative, Neutral, Joyful, Angry, Sad, Surprised, Confident, Anxious, Confused

Compare: Chat Completions

Run the same prompt against multiple LLM models in a single call. Returns one choice per model. Optionally saves the result as a shareable snapshot.

import requests, os

response = requests.post(
    "https://app.gmtech.com/v1/compare/completions",
    headers={"Authorization": f"Bearer {os.environ['GMTECH_API_KEY']}"},
    json={
        "models": ["gpt-4o", "claude-sonnet-46", "google-gemini-25-pro"],
        "messages": [{"role": "user", "content": "Explain quantum entanglement in one sentence."}],
        "gmtech": {"create_snapshot": True, "snapshot_public": True}
    }
)

data = response.json()
for choice in data["choices"]:
    print(choice["model"], "→", choice["message"]["content"])

print(data["gmtech"]["snapshot_url"])  # shareable link

Parameters

Response

Returns choices — one entry per model — plus a gmtech block if a snapshot was created:

{
  "object": "chat.compare.completion",
  "choices": [
    {
      "index": 0,
      "model": "gpt-4o",
      "message": {"role": "assistant", "content": "..."},
      "finish_reason": "stop",
      "usage": {"prompt_tokens": 20, "completion_tokens": 30, "total_tokens": 50},
      "gmtech": {"cost_usd": 0.000312}
    }
  ],
  "gmtech": {
    "snapshot_uuid": "abc123...",
    "snapshot_url": "https://app.gmtech.com/snapshots/abc123..."
  }
}

Compare: Image Generation

Run the same image prompt against multiple image models in a single call. Returns one image URL per model. Optionally saves the result as a shareable snapshot.

import requests, os

response = requests.post(
    "https://app.gmtech.com/v1/compare/images",
    headers={"Authorization": f"Bearer {os.environ['GMTECH_API_KEY']}"},
    json={
        "models": ["gpt-image-1", "google-imagegen-4"],
        "prompt": "A futuristic city at sunset",
        "gmtech": {"create_snapshot": True, "snapshot_public": True}
    }
)

data = response.json()
for choice in data["choices"]:
    print(choice["model"], "→", choice["url"])

print(data["gmtech"]["snapshot_url"])  # shareable link

Parameters

Response

{
  "object": "image.compare.completion",
  "choices": [
    {
      "index": 0,
      "model": "gpt-image-1",
      "url": "https://storage.googleapis.com/.../abc123.webp",
      "finish_reason": "stop",
      "gmtech": {"cost_usd": 0.042}
    }
  ],
  "gmtech": {
    "snapshot_uuid": "abc123...",
    "snapshot_url": "https://app.gmtech.com/snapshots/abc123..."
  }
}

Available Models

Use any model key from the GMTech model directory as the model parameter.

Browse all models →  |  LLM & embedding model keys (API) →  |  Image model keys (API) →

JavaScript/TypeScript

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://app.gmtech.com/v1',
  apiKey: process.env.GMTECH_API_KEY
});

const response = await client.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello!' }]
});

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

cURL

curl https://app.gmtech.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Api-Key gmtech_your_api_key" \
  -d '{
    "model": "gpt-4o",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Supported Features

Models & pricing

GET /v1/models lists every active model (OpenAI-compatible) and adds a gmtech block per model with its kind and per-unit pricing. Filter by kind with ?type=llm, ?type=ig (image), or ?type=vid (video).

models = client.models.list()           # OpenAI SDK
# or: GET https://app.gmtech.com/v1/models?type=llm

{
  "object": "list",
  "data": [{
    "id": "gpt-4o",
    "object": "model",
    "owned_by": "openai",
    "gmtech": {
      "type": "llm",                  # llm | ig | vid
      "price_input": 0.0025,          # per pricing_unit
      "price_output": 0.01,
      "price_flat": null,              # set for per-image/per-video models
      "pricing_unit": "per_1k_tokens",
      "tags": ["vision", "best"]
    }
  }]
}

A dedicated GET /api/v1/pricing/models/ (public, no auth) returns the same pricing data plus company/product/shortname — handy for a pricing page or cost estimates. Same ?type= filter. Pricing is straight-through: what we pay the provider is what you pay.

Every generation response also reports what it cost under a gmtech block: chat, images, and embeddings all return gmtech.cost_usd (the exact amount debited from your balance); chat and embeddings also return OpenAI-standard usage token counts.

Vision (image input)

Vision-capable models accept images alongside text using the standard OpenAI multimodal content format. Pass a data: URI (base64) or a public https image URL. Sending an image to a model that isn't vision-capable returns a 400. To see which models support vision, filter by the vision tag on the model status page.

import base64
from openai import OpenAI

client = OpenAI(base_url="https://app.gmtech.com/v1", api_key="YOUR_KEY")

# Option A — a public image URL
resp = client.chat.completions.create(
    model="gpt-4o",  # any model tagged "vision" (gpt-4o, gemini-2.5-pro, claude-*, llama-4-*, grok-4.x, nova-*, ...)
    messages=[{"role": "user", "content": [
        {"type": "text", "text": "What is in this image?"},
        {"type": "image_url", "image_url": {"url": "https://example.com/photo.jpg"}},
    ]}],
)
print(resp.choices[0].message.content)

# Option B — inline base64 (data URI)
b64 = base64.b64encode(open("photo.jpg", "rb").read()).decode()
resp = client.chat.completions.create(
    model="gemini-2.5-pro",
    messages=[{"role": "user", "content": [
        {"type": "text", "text": "Describe this."},
        {"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{b64}"}},
    ]}],
)

Notes: images are billed via input tokens (larger images = more tokens; the provider counts them). Remote image URLs must be public https (fetched with SSRF protections; redirects and private/internal hosts are rejected). Images on the last user message are sent to the model.

Internet search

Add internet_enabled: true to run a live web search, inject the results into your prompt, and return citations. Off by default. A per-search fee (search_cost_usd) is charged on top of token cost; sources are returned under gmtech.sources.

resp = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "What happened in AI this week?"}],
    extra_body={"internet_enabled": True},  # default False
)
print(resp.choices[0].message.content)
print(resp.model_extra["gmtech"]["sources"])  # [{n, title, url}, ...]

Error Handling

from openai import APIError, AuthenticationError, RateLimitError

try:
    response = client.chat.completions.create(
        model="gpt-4o",
        messages=[{"role": "user", "content": "Hello!"}]
    )
except AuthenticationError:
    print("Invalid API key")
except RateLimitError:
    print("Rate limit exceeded")
except APIError as e:
    print(f"API error: {e}")

Migration from OpenAI

Before (OpenAI)

from openai import OpenAI

client = OpenAI(api_key="sk-...")
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}]
)

After (GMTech)

from openai import OpenAI

client = OpenAI(
    base_url="https://app.gmtech.com/v1",  # Only change
    api_key="gmtech_..."                 # Only change
)
response = client.chat.completions.create(
    model="gpt-4o",  # Can also use claude-sonnet-46, google-gemini-25-pro, etc.
    messages=[{"role": "user", "content": "Hello"}]
)

Next Steps

• Code Examples - More examples in Python, JavaScript, cURL
• Authentication - API key management and security
• Overview - Complete API reference

Need help? Contact support or join office hours