Step 1
Recharge API balance in USD
Open Pricing, choose the API tab, and add API balance from $9 to $200. The balance is stored separately from website generation credits.
Kimi K3 API
OpenAI-compatible Kimi K3 access for apps, agents, and automation. Use chat completions, streaming, tool calling, JSON outputs, image input, user API keys, and dollar-based billing.
OpenAI compatibleKimi K3 modelStreaming SSETool callingJSON response_formatImage inputUser API keysDollar billing
API usage
Start with this Kimi K3 API request
Use your user API key as a Bearer token. The primary endpoint is OpenAI-compatible and accepts the model id kimi-k3.
Base URL
https://api.viddiai.com/v1
Chat Completions
POST /v1/chat/completions
Model
kimi-k3
API key
Create it in Settings > Security
Minimal curl example
curl https://api.viddiai.com/v1/chat/completions \
-H "Authorization: Bearer wk_your_user_api_key" \
-H "Content-Type: application/json" \
-d '{
"model": "kimi-k3",
"messages": [
{"role": "user", "content": "Write a short product launch email."}
]
}'OpenAI SDK example
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.KIMI_API_KEY,
baseURL: "https://api.viddiai.com/v1"
});
const completion = await client.chat.completions.create({
model: "kimi-k3",
messages: [
{ role: "user", content: "Explain Kimi K3 API billing in one paragraph." }
]
});
console.log(completion.choices[0]?.message?.content);Token pricing
Kimi K3 API pricing is built for high-volume agent work
Pay from your API balance only when tokens are used. Kimi K3 is priced at roughly 1/2.2 of Fable 5 while targeting performance closer to Fable 5 for coding, agent, and reasoning-heavy workflows.
Closer to Fable 5 performance, priced for everyday usage
Use Kimi K3 for code generation, tool-calling agents, JSON extraction, long-form drafting, and multimodal prompts when you want premium-model quality without premium-model token cost.
Usage
Kimi K3
vs Fable 5
Usage
Uncached input
Kimi K3
$4.50 / 1M tokens
vs Fable 5
About 1/2.2 of Fable 5
Usage
Cached input
Kimi K3
$0.45 / 1M tokens
vs Fable 5
About 1/2.2 of Fable 5 cached input
Usage
Output
Kimi K3
$22.50 / 1M tokens
vs Fable 5
About 1/2.2 of Fable 5
Fable 5 reference pricing: $10 input and $50 output per 1M tokens.
API recharge is dollar balance. There is no credits conversion.
What is the Kimi K3 API
A production-ready Kimi K3 API relay with accounts, recharge, usage tracking, and OpenAI-compatible requests
The Kimi K3 API gives developers a simple gateway for Kimi K3. Instead of sharing one provider key, every user can create their own API key, recharge a separate API dollar balance, and call an OpenAI-compatible endpoint.
API balance is separate from website generation credits. Chat requests are billed from the API balance in dollars, so subscriptions, yearly credits, one-time web credits, and API usage do not get mixed together.
How to use the Kimi K3 API
Create an account, add API balance, generate a key, then call the OpenAI-compatible endpoint from your own app.
The fastest setup is: recharge balance first, create a key second, then test with curl or the OpenAI SDK.
Step 2
Generate your user API key
Go to Settings > Security, create a Kimi K3 API key, and copy it immediately. The full key is shown only once. Use it as a Bearer token in every request.
Step 3
Call the OpenAI-compatible endpoint
Send requests to https://api.viddiai.com/v1/chat/completions with model kimi-k3. The alternate path https://api.viddiai.com/api/v1/chat/completions is also supported.
Step 4
Track usage and balance
Open Settings > Profile to see current API balance, total recharged, total used, request count, and recent API activity. Settings > Security also shows API keys and recent balance events.
API recharge
Add Kimi K3 API balance in USD
Recharge from $9 to $200 and use the balance only for Kimi K3 API calls. Website generation credits and API spending stay separate.
API balance
$9
Use until this API balance is spent.
API balance
$20
Use until this API balance is spent.
API balance
$50
Use until this API balance is spent.
API balance
$100
Use until this API balance is spent.
API balance
$200
Use until this API balance is spent.
Kimi K3 API features
Everything needed for a basic API relay product: keys, balance, billing, compatibility, streaming, and account visibility.
The Kimi K3 API gateway is designed for developers who need OpenAI-style requests with per-user access control and transparent dollar billing.
01
OpenAI-compatible Chat Completions
Use /v1/chat/completions with the same request shape developers already know from OpenAI-compatible clients.
02
Realtime streaming
Streaming responses are forwarded as SSE chunks as they arrive. The relay captures the final usage chunk and charges after the stream finishes.
03
Tool calling support
Function-style tool definitions and tool call responses are supported, making the API usable for agents, workflows, and structured app logic.
04
Structured JSON outputs
Use response_format to request JSON object responses for extraction, classification, routing, and application data generation.
05
Image input compatibility
Base64 data image input is supported. Public image URLs are fetched by the gateway and converted before being sent upstream.
06
Separate API dollar balance
API usage is charged from API balance, not website credits. This keeps subscription credits and API spending cleanly separated.
Kimi K3 API examples
Use these request patterns for the most common Kimi K3 API jobs.
Streaming with usage
Set stream to true. The gateway requests stream_options.include_usage from upstream so the final usage chunk can be used for exact billing.
Endpoint: POST /v1/chat/completions
Request option: stream: true
Billing: charged after final usage is received
Streaming with usage
const stream = await client.chat.completions.create({
model: "kimi-k3",
stream: true,
stream_options: { include_usage: true },
messages: [{ role: "user", content: "Count from 1 to 5." }]
});
for await (const chunk of stream) {
process.stdout.write(chunk.choices[0]?.delta?.content || "");
}Tool calling
Pass OpenAI-style tools and let Kimi K3 return tool_calls when it decides an external function should be used.
Use tools with function schemas
Read choices[0].message.tool_calls
Send the tool result back in the next turn
Tool calling
const completion = await client.chat.completions.create({
model: "kimi-k3",
messages: [{ role: "user", content: "What is the weather in Tokyo?" }],
tools: [{
type: "function",
function: {
name: "get_weather",
description: "Get weather for a city",
parameters: {
type: "object",
properties: { city: { type: "string" } },
required: ["city"]
}
}
}]
});JSON response_format
Use response_format when your application needs parseable JSON instead of prose.
Best for extraction and classification
Use response_format: { type: "json_object" }
Validate the returned JSON in your application
JSON response_format
const completion = await client.chat.completions.create({
model: "kimi-k3",
response_format: { type: "json_object" },
messages: [{
role: "user",
content: "Return JSON with keys ok and summary."
}]
});Image input
Send OpenAI-style image_url content parts. Base64 data URLs are supported directly. Public image URLs are fetched and converted by the gateway.
Use content parts with type image_url
Base64 data URLs work directly
Public image URLs are limited by the gateway fetch size
Image input
const completion = await client.chat.completions.create({
model: "kimi-k3",
messages: [{
role: "user",
content: [
{ type: "text", text: "Describe this image in one sentence." },
{ type: "image_url", image_url: { url: "https://example.com/image.jpg" } }
]
}]
});Billing and account rules
API recharge is a simple dollar balance. Users recharge, call the API, and pay only for actual Kimi K3 usage.
01
Recharge range
API balance packages currently start at $9 and go up to $200. Users can recharge from the API tab on the Pricing page.
02
Input token rate
Kimi K3 input billing is $4.50 per 1M uncached input tokens.
03
Cached input rate
Cached input billing is $0.45 per 1M cached input tokens.
04
Output token rate
Output billing is $22.50 per 1M output tokens.
05
Exact usage calculation
The gateway reads upstream usage, separates cached and uncached prompt tokens when available, adds output tokens, calculates the exact dollar cost, and writes one usage transaction.
06
Balance visibility
Users can see current balance, total recharged, total used, API request count, and recent transactions in Settings > Profile.
Kimi K3 API FAQ
Everything developers need to know before using the Kimi K3 API gateway.
What is the Kimi K3 API endpoint?
Use https://api.viddiai.com/v1/chat/completions. The alternate path https://api.viddiai.com/api/v1/chat/completions is also supported.
Which model id should I use?
Use kimi-k3. The gateway also accepts a few common aliases and normalizes them to kimi-k3 upstream.
Where do users create API keys?
Users create API keys in Settings > Security. The full key is shown only once, and only a hash is stored.
How does API recharge work?
Users recharge a separate API dollar balance from the API tab on Pricing. This balance is not the same as website generation credits.
How is streaming billed?
Streaming responses are forwarded in realtime. After the final usage chunk arrives, the gateway calculates the exact cost and records the usage transaction asynchronously.
Does the API support tool calling?
Yes. OpenAI-style tools are supported and Kimi K3 can return tool_calls in the assistant message.
Does the API support response_format?
Yes. response_format with JSON object output has been tested and is supported for structured responses.
Does image input work?
Yes. Base64 data image input works directly. Public image URLs are fetched and converted by the gateway before calling upstream.
Build with Kimi K3
Use Kimi K3 from your app with a user API key and exact dollar billing
Recharge API balance, create a key, and call the OpenAI-compatible Kimi K3 endpoint from your backend, agent, workflow, or product.