KEDA / API
THE API FOR YOUR NEXT IDEA

One API.
More possibilities.

Connect leading models through one API. Combine channels and switch models, giving every request another way forward.

Create APIExplore smart routing
  • Works with existing SDKs
  • Many channels. One key.
  • Pay for actual usage
hello_keda.py
from openai import OpenAI
import os

client = OpenAI(
    api_key=os.environ["KEDA_API_KEY"],
    base_url="https://keda8.ai/v1",
)

response = client.chat.completions.create(
    model="YOUR_CONFIGURED_MODEL_ID",
    messages=[{
        "role": "user",
        "content": "Make something great."
    }],
)

print(response.choices[0].message.content)
Your code. More models.

Replace model with a full model ID available to your key.

More possibilities for your next project
OpenAI logoOpenAIClaude logoClaudeGemini logoGeminiDeepSeek logoDeepSeekGrok logoGrok
Explore available models
ONE KEY. MORE POSSIBILITIES.

Connect once. Build more.

From one model to many channels. Let the API handle connections so you can focus on your product.

01

Combine your models

One integration. More models.

Access the combined models of your selected channels. Choose the right model for conversation, coding or image understanding.

02

Control your routing

One request. Multiple routes.

Choose channels and priorities for your application, matching each request to a model with the capabilities it needs.

03

Plan for interruptions

Switch channels. Keep going.

When your primary channel fails, try other channels, then fallback models in your chosen order.

COMPOSE YOUR API

Your API. Your configuration.

Use the same live channels and configuration as the console. Sign in to review and create.

FROM IDEA TO FIRST REQUEST

From idea to your first request.

Keep your familiar workflow and connect to more models.

Read the integration guide
  1. 01

    Create an API

    Name your application, select available channels, and set quota and model access.

  2. 02

    Connect your application

    Set your Base URL and API key, then specify a full model ID in your request.

  3. 03

    Start building

    Send your first request and view actual usage and logs in the console.

GOOD TO KNOW

A few more things to know.

Can one API access models from different providers?

Yes. Select multiple available channels for a key to access their combined models. Specify a model on each request. Model access and balances are checked by the server.

How do same-model retries and cross-model fallback differ?

Same-model retries keep the model ID and try other channels. Smart routing falls back to other models in your chosen order. Cross-model fallback supports Chat Completions and stops once output begins.

Does selecting multiple channels call all models at once?

Each request follows one available route. The same API can call different models across requests. Calling several models in parallel requires separate orchestration.

Do all models accept the same parameters?

Endpoints, tools, vision and context support vary by model. Confirm compatibility before enabling fallback; see each model for details.

BUILD SOMETHING THAT MATTERS

Build what comes next.