AI API for Latin American teams

China-developed and global AI models. One OpenAI-compatible API.

NexoRouter helps developers, startups, and product teams test DeepSeek, Qwen, Kimi, GLM, and more through one integration, with published routes, prepaid balance, and usage controls.

Create account and get API keyCompare models
OpenAI-compatible
One API
Models for each task
More options
Costs under control
Prepaid
Keys, balance & logs
Visible

Quick answer

What does a unified AI API solve?

A unified AI API lets a team integrate several models through the OpenAI format, choose the right option for each use case, and operate credentials and spend from one workspace. For Latin American teams, it lowers the cost of switching providers and makes usage more predictable with prepaid balance and logs.

China-developed AI models for Latin America

More options to build, compare, and control the cost of your AI product.

China-developed models add options for coding, agents, reasoning, and multimodal work. NexoRouter brings them together with global models behind an OpenAI-compatible API, so you can evaluate quality, context, capabilities, and published pricing before sending production traffic.

Explore China AI models
72
reviewed models in this group
8
verified families
01

Change models without rebuilding the client

Keep an OpenAI-compatible interface and decide by task, latency, context, or budget instead of committing to one provider.

02

Visible costs before you scale

Review published token rates and use prepaid balance to plan tests, pilots, and production workloads with a clear operating reference.

03

A broader selection for agents and products

Evaluate families with distinct capabilities—reasoning, coding, tools, vision, or long context—through the same base URL and workspace controls.

Practical infrastructure

From model trial to an operation your team can review.

Models, capabilities, and prices

A catalog built for real comparison

Review provider, capability, identifier, availability, and published price before choosing a route.

OpenAI-compatible base URL

One integration for your stack

Connect OpenAI SDK, Vercel AI SDK, LangChain, or HTTP without introducing another proprietary interface.

Logs and operational signals

Requests your team can review

Inspect activity, status, and errors to debug an integration and decide with context.

Keys, balance, and permissions

Every project under control

Create separate keys, add prepaid balance, and avoid mixing credentials between products or teams.

Choose for the job, not the brand

Models for AI agents, coding, and multimodal products.

01

Code and agents

Compare reasoning, tool calling, structured JSON, and context capacity for automations and copilots.

Tool callingJSONContext

02

Multimodal product

Explore text, vision, audio, or image models according to the published requirements for each route.

TextVisionAudio

03

OpenAI migration

Keep your SDK and change the base URL to test alternatives without redesigning the whole integration layer.

Same SDKNew URL

04

Budgeted testing

Separate API keys by project, add balance, and use logs to observe consumption and errors before you scale.

BalanceKeysLogs

Explore the catalog from here

Text, image, video, and audio in the same directory

View all
247
published models
25
providers
6
capability types
Selection refreshed dailyLast sync: Sep 10, 2026, 7:19 PM

Today’s featured China-developed AI models

This selection changes with newly verified routes, recent usage signals, and each family’s current generation. We only show available models with published pricing.

View all
Kimi
Popular nowChat

Moonshot AI

Kimi K3

Input
$3.04 / 1M tokens
Output
$15.20 / 1M tokens
Open profile
Z.ai
Popular nowChat

Zhipu AI

GLM 5.1

Input
$1.22 / 1M tokens
Output
$4.26 / 1M tokens
Open profile
DeepSeek
Popular nowChat

DeepSeek

DeepSeek V4 Flash 0731

Input
$0.46 / 1M tokens
Output
$1.37 / 1M tokens
Open profile
ByteDance
Popular nowImage

ByteDance

Doubao Seedream 5 0 Pro 260628

Input
$0.09 / image
Output
Included
Open profile

Works with your team’s stack

Use the NexoRouter base URL with tools that already speak the OpenAI format.

View all

OpenAI SDK

Python and JavaScript

Existing code, new base URL.

Vercel AI SDK

Streaming and UI

Connect web apps with one key.

LangChain

Chains and agents

Keep a familiar interface.

HTTP / cURL

Direct control

Integrate from any platform.

From zero to your first request

Three steps, with a clear result at each one.

The catalog is public. When you are ready to test a route, create your workspace, load a controlled budget, and generate a key for your project.

  1. 01

    Your workspace

    Create your account

    Open a personal workspace with access to the console and your projects.

    Create account
  2. 02

    Prepaid budget

    Add balance

    Top up with Stripe and confirm available credit before sending traffic.

    See pricing
  3. 03

    One key per project

    Generate an API key

    Copy the base URL, choose a model, and make your first request.

    Read the guide
Your first requestPOST /v1/chat/completions
curl https://api.nexorouter.com/v1/chat/completions \
  -H "Authorization: Bearer $NEXOROUTER_API_KEY" \
  -d '{ "model": "your-model-id", "messages": [...] }'

Same format, more models

Learn and build

From your first call to a production-ready integration.

Practical tutorials, technical documentation, and operating guides connected to NexoRouter’s real capabilities.

Explore all resources
Featured tutorial

Tutorial

Build your first integration

Follow a reproducible cURL and JavaScript example using the current catalog.

client = OpenAI(
  base_url="https://api.nexorouter.com/v1",
  api_key=os.environ["NEXOROUTER_API_KEY"],
)

models = client.models.list()

Docs

Getting started

Configure the base URL, authenticate your key, and complete a test request.

Model selection

China and global models

Evaluate DeepSeek, Qwen, Kimi, and global alternatives with the same criteria.

Operations

Cost controls and logs

Set up dedicated keys, budgets, output limits, and request traceability.

Agents

Agents and tools

Connect an agent without exposing credentials or losing operational control.

ReferenceAPI referenceReview endpoints, parameters, responses, and errors before going to production.UpdatesProduct and API changesReview published changes across the catalog, documentation, and platform.

Frequently asked questions

The essentials before you create an API key.

What is NexoRouter?

NexoRouter is a platform for accessing AI models through an OpenAI-compatible API. It centralizes the catalog, API keys, prepaid balance, and usage logs in one workspace.

Can I use code based on OpenAI?

Yes. NexoRouter keeps an OpenAI-compatible interface so you can update the base URL and API key, then test published models without rebuilding the entire integration.

Why evaluate China-developed AI models?

Families such as DeepSeek, Qwen, Kimi, and GLM provide options for different coding, reasoning, tool, context, and multimodal tasks. The right choice should be compared against each route’s published requirements and rates.

How do I control API spending?

You can use prepaid balance, create API keys per project, and inspect usage logs. This provides a clear consumption reference before increasing traffic.

One integration. More room to build.

Create your API key, compare models, and validate your first use case with the published catalog.

Create workspace