What is Kyma API?

Overview

Kyma API is an LLM API gateway that routes one key to 106 models and publishes measured uptime per model. It gives you instant access to the best open source and frontier LLMs through a single endpoint. Compatible with both OpenAI and Anthropic SDKs. Many models have more than one route, so a request can fail over when one route is down; measured uptime for every model is on Status.

106 models, one endpoint

Qwen 3.6, DeepSeek V4, Gemma 4, GPT-OSS, Kimi K2.6, Gemini, Llama, MiniMax, GLM, Perplexity Sonar for live web search, plus GPT Image 2, FLUX, Ideogram, Recraft, MiniMax Image for image generation, ElevenLabs and MiniMax for voice.

Auto-Failover

Where a model has more than one route, a failed request is retried on another route automatically.

OpenAI Compatible

Drop-in replacement. Works with any OpenAI SDK, LangChain, Cursor, and more.

Why Kyma?

	Kyma	Other gateways	Direct API
Price	$0.50 free credit after email verification, free tier	$0.001+/token	Free (limited)
Models	Curated active models	200+ (paid)	1 per provider
Redundancy	Auto-failover	Single provider	None
Setup time	30 seconds	5 minutes	Per-API signup
OpenAI SDK	✅	✅	✅

The

0.50 signup credit is added when you verify your email (Google sign-in gets it at once), and it spends on the free tier: {/* GENERATED:free-tier-thresholds */}text models at or under

3/M output and

0.50/M input, plus media at or under

0.01 per billing unit. Most image models and every text-to-speech model need a top-up first; minimax-image-01 is the free-tier image model. Music and realtime sessions are paused. See Free tier. Updated 2026-09-22.

How it works

Your App → Kyma API → a route that serves that model
                    ↳ another route for the same model, if the model has one
                    ↳ otherwise an error, or a substitute model named in x-kyma-served-model

Many models have more than one route, and Kyma tries them in order of live health, so a single route going down does not reach you. A model with one route has nothing to fail over to: when it is down you get a 5xx, or an answer from a substitute model. A substitute is always named in the x-kyma-served-model response header, so compare it with the model you asked for. Measured uptime per model is on Status.

What happens to a request

The checks run in this order, so a request refused early costs nothing. usage.cost is the price of the call; with stream: true it arrives on the final usage chunk. Your balance is charged after the response ends. What happens when a route fails is in Routing.

Quick example

from openai import OpenAI

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

response = client.chat.completions.create(
    model="llama-3.3-70b",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://kymaapi.com/v1",
  apiKey: "YOUR_KEY",
});

const response = await client.chat.completions.create({
  model: "llama-3.3-70b",
  messages: [{ role: "user", content: "Hello!" }],
});
console.log(response.choices[0].message.content);

curl https://kymaapi.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama-3.3-70b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Ready to start?

Get your API key and make your first request in 30 seconds →

Getting Started

More

What is Kyma API?

Overview

106 models, one endpoint

Auto-Failover

OpenAI Compatible

Why Kyma?

How it works

What happens to a request

Quick example

Ready to start?

​Overview

106 models, one endpoint

Auto-Failover

OpenAI Compatible

​Why Kyma?

​How it works

​What happens to a request

​Quick example

Ready to start?

Overview

Why Kyma?

How it works

What happens to a request

Quick example