xKiro

About xKiro

Where agentsmeet models.

xKiro is an AI gateway for developers and AI agents. One API key and one endpoint reach 103 models from 14 providers — through the OpenAI and Anthropic SDKs you already use.

Your code
  • OpenAI SDK/v1/chat/completions
  • Anthropic SDK/v1/messages
  • Claude Code/v1/messages
  • Codex CLI/v1/responses
  • Cline/v1/chat/completions
  • curl/v1/chat/completions
api.xkiro.com

https://api.xkiro.com/v1

One key · one endpoint

Models
  • OpenAI8 models
  • Anthropic15 models
  • Google8 models
  • DeepSeek3 models
  • Qwen19 models
  • Grok3 models
  • Kimi2 models
+7 more providersBrowse all 103 models

Each client is shown with the endpoint it calls. Providers and model counts are read from the live catalog.

New model? Change one string.

All 103 models sit behind the same endpoint. Switching is a one-line diff.

  • Same API key
  • Same endpoint
  • Same SDK
agent.ts — the only line that changes
const res = await client.chat.completions.create({  model: "anthropic/claude-opus-5.5",  model: "openai/gpt-6.1-sol",  messages,});

Spec sheet

xKiro, without the adjectives.

The model numbers on this sheet are counted from the live catalog and refreshed every hour. None of them is typed in by hand.

Endpoint
https://api.xkiro.com/v1
Works with
OpenAI SDK · Anthropic SDK · any HTTP client
Quickstart
Models
103from 14 providers
Browse the catalog
Free to call
45models · no credit card
Longest context
1.05Mtokens in a single request
APIs
  • /v1/chat/completions
  • /v1/messages
  • /v1/responses
  • /v1/images/generations
  • /v1/audio/speech
  • /v1/search
  • /v1/fetch
  • /mcp/web
API reference
This site
6languagesEnglish · Tiếng Việt · 简体中文 · Русский · Português (Brasil) · Bahasa Indonesia
Support
Tiếng Việt · English
Contact the team

Counted live from the model catalog · updated hourly

How we work

Four rules, each one in writing.

Every rule below links to the page of our documentation that commits to it — so you can hold us to it.

  1. 01

    Compatible by default

    Keep the SDK, the coding tool and the code you already have. Point the base URL at xKiro and nothing else has to change.

    baseURL
    "https://api.xkiro.com/v1"
    apiKey
    process.env.XKIRO_API_KEY
    model
    "anthropic/claude-opus-5.5"
    Quickstart
  2. 02

    Counted, not claimed

    Every model count and price on this site is read from the live catalog — each model lists its input and output price per million tokens before you call it.

    anthropic/claude-opus-5.5
    $4 / $20
    openai/gpt-6.1-sol
    $2 / $10
    deepseek/deepseek-v4-pro-0813
    $0.2828 / $0.5655
    Model catalog
  3. 03

    Pay for what reached you

    A request that fails with a 5xx is not billed. Cancel a stream and you pay only for the tokens already generated. Send the same blocking request again by accident and it is replayed, not billed twice.

    5xx
    not billed
    stream cancelled
    tokens generated so far
    same request again
    replayed, billed once
    Pricing & billing
  4. 04

    Your keys, your limits

    A key is shown once and stored only as a hash. Give each service its own key with its own monthly spending limit, and revoke any key at any time.

    sk-xt-…
    shown once, stored as a hash
    limit
    monthly, per key
    revoke
    any time, from the console
    Authentication

Start with one key.

Create an API key, point your SDK at xKiro and send your first request — 45 models are free to call.

Get your API keyTalk to the team

No credit card · Cancel anytime · 45 free models, free forever