Explore multimodal AI.
Build with one API.

Try multimodal models in the playground, then use them in your applications. We handle the hosting.

Open console Read the docs
api.reka.aiauth + meteringprivate clouddedicated capacitytokens
The gateway decides whether you are on shared or dedicated capacity, then your request stays on a chassis we own until the tokens leave.

Make your first API call

Create an account with the link at the top of this page, or sign in if you already have one. Then create an API key and save it as REKA_API_KEY in your environment. Choose a model and run the example below.

One endpoint. Your choice of model.

Start with DeepSeek in this example, or change model to an ID from the model list.

curl https://api.reka.ai/v1/chat/completions \
  -H "Authorization: Bearer $REKA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek4-flash",
    "messages": [
      { "role": "user", "content": "Why is the sky blue?" }
    ]
  }'

Use your existing OpenAI client

Set the base URL and API key, then specify the model you want to call.

npm install openai
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.reka.ai/v1",
  apiKey: process.env.REKA_API_KEY,
});

const res = await client.chat.completions.create({
  model: "deepseek4-flash",
  messages: [{ role: "user", content: "Why is the sky blue?" }],
});

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

See the API reference for streaming, supported parameters and error codes.

Models

Different tasks call for different models. Choose from popular open models and Reka models, all served on infrastructure we operate. Browse the model catalog for descriptions and capabilities.

Text model prices per million tokens, from /v1/models as of 5 October 2026.

ModelContextInput $/MOutput $/MCached input $/M
deepseek4-flash1M$0.11$0.66$0.007
deepseek-v4-pro1M$1.20$3.30$0.13
glm5.3262k$1.17$3.96$0.234
glm5.3-flash262k$0.15$0.50$0.03
qwen3.8-27b262k$0.20$2.50$0.05
reka-edge-260316k$0.10$0.10—
reka-flash-364k$0.10$0.20—

Clear, usage-based pricing

Pay for the tokens you use at the rates above, with no platform fee or markup. Where prompt caching is supported, cached input is billed at the model’s cached-input rate.

Compare input, output and cached-input rates for your workload in the model catalog.

Add credit and track spending on the usage page.

Built and operated by Reka

We build models and run inference. Reka Gateway brings our models and a curated selection of open models together through one API. You choose the model for each request; we run it on infrastructure we operate.

The gateway authenticates your request and tracks usage. Our team operates the serving infrastructure behind it, from the API to the GPUs.

Need dedicated inference?

We also offer dedicated capacity. Talk to our team about your models, workload and deployment needs.

Contact Reka