One Chinese Model API
One key reaches DeepSeek, Kimi, Qwen, GLM, Doubao, MiniMax and more without repeated vendor integrations.
New Sign-ups Get $5 Free Credit200 OK
The OpenRouter for Chinese frontier AI Models
Access DeepSeek, Kimi, Qwen, GLM, Doubao and MiniMax through one OpenAI-compatible API built for global developers evaluating Chinese model families.
OpenAI-compatible console
01Your app
02SmarToken gateway
03China model route
Chat Completions
Change only baseURL and model.
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://thesmartoken.com/v1",
apiKey: process.env.SMARTOKEN_API_KEY
});
await client.chat.completions.create({
model: "deepseek-chat",
messages: [{ role: "user", content: "Build an agent plan" }],
stream: true
});Available routes
DeepSeek
Kimi
Qwen
GLM
Budgets, routing and usage stay visible in one console.
One API for global teams choosing Chinese frontier models
Chinese model access without vendor-by-vendor integration
China-first Routing
Route across Chinese model families by use case, cost and availability instead of treating them as a generic catalog.
OpenRouter Migration
Keep OpenAI-compatible SDKs and use familiar model slugs while moving China-model traffic to SmarToken.
Why Chinese models
A focused alternative when your priority is DeepSeek, Qwen, Kimi and GLM
Broad aggregators are useful when every model family matters. SmarToken optimizes the narrower job: help overseas builders discover, test and route Chinese frontier models through a familiar API.
Cost-efficient coding and agent workloads
DeepSeek, Qwen, MiniMax and Kimi-style routes give teams more options for code generation, agent planning and high-volume automation.
Long context and multilingual coverage
Chinese frontier models are increasingly strong for long-document work, bilingual products and structured output workflows.
Redundancy beyond Western providers
A China-focused gateway gives global teams a second model supply line for evaluation, fallback and price negotiation.
Service flow
From signup credit to auditable model usage
The practical path is simple: create a key, choose a Chinese model by use case, call the OpenAI-compatible endpoint, then inspect tokens, cost and latency before scaling production traffic.
Step 1
Sign in and receive trial credit
$5
trial credit
Google login creates a console account. Eligible new accounts receive promotional USD credit before topping up.
Step 2
Create a scoped API key
sk
console keys
Generate sk keys with model-family scope, daily budgets and monthly budgets for each app or agent.
Step 3
Call model routes
/v1
API base
Use the OpenAI SDK with SmarToken baseURL and choose native IDs or OpenRouter-style aliases for China-first models.
Step 4
Review usage and billing
log
traceability
Track prompt tokens, output tokens, estimated USD cost, latency, top-ups and wallet ledger entries from the dashboard.
China model entry points
Canonical API pages for Chinese frontier models
DeepSeek
Reasoning and coding
/models/deepseek-chat
Kimi
Long-context agents
/models/kimi-k2
Qwen
Multilingual apps
/models/qwen-plus
GLM
Enterprise assistants
/models/glm-4-flash
Doubao
Consumer AI workflows
/models/doubao-seed
ERNIE
Chinese knowledge QA
/models/ernie-4-turbo
Hunyuan
Tencent Cloud routes
/models/hunyuan-turbos
MiniMax
Coding agents
/models/minimax-m2
StepFun
Planning workflows
/models/step-3
Spark
Chinese reasoning
/models/spark-x1
Pangu
Industry NLP
/models/pangu-nlp
LongCat
Open model evals
/models/longcat-flash
Bring Chinese AI capability
Make Chinese AI easier to use
Support global developers with Chinese AI technologies, no matter where their business runs.
Available globally
Accelerate the Arrival of AGI
For global developers
North AmericaEuropeSoutheast AsiaMiddle EastLatin AmericaIndia
OpenAI Compatible
Use familiar Chat Completions, bearer auth, SDKs and streaming patterns.
More Stable Availability
Multi-route infrastructure improves continuity for cross-border model access.
Built for Overseas Teams
English docs, global onboarding and console workflows for international builders.
Trust and scope
Clear boundaries before teams route production traffic
Overseas developers need to know which model family they are calling, how spend is controlled and how existing OpenRouter-style integrations can move to a China-model-focused gateway.
SmarToken is an API gateway for Chinese frontier models, not a broad consumer chat product.
Prepaid balance, Stripe, Alipay, WeChat Pay and USDT top-up paths are separated from upstream vendor keys.
Budget limits stop over-spend before provider calls when a key is revoked, scoped out or over budget.
OpenRouter-style model aliases reduce migration friction for teams already using aggregator APIs.
Docs, model pages, FAQ, support tickets, announcements and billing records remain reachable from the product shell.
Developer quick answers
Short answers for developers comparing Chinese AI APIs
What is SmarToken?
SmarToken is an OpenAI-compatible API gateway for overseas developers who need one key to access Chinese AI models such as DeepSeek, Kimi, Qwen, GLM, Doubao, ERNIE, Hunyuan, MiniMax, Spark, Pangu and more.
How do I call Chinese AI models with the OpenAI SDK?
Keep the OpenAI SDK, set baseURL to https://thesmartoken.com/v1, authenticate with a SmarToken key, and choose a supported model ID such as deepseek-chat, kimi-k2 or qwen-plus.
Which Chinese model should I start with?
Start with DeepSeek for coding and cost-effective reasoning, Kimi for long-context work, Qwen for multilingual and structured output, MiniMax for coding agents, Hunyuan for Tencent routes, Spark for Chinese reasoning, and ERNIE or GLM for enterprise tasks.
Why use SmarToken instead of direct vendor accounts?
SmarToken combines English docs, Playground testing, budget-limited API keys, usage logs and route-pool management so global teams can evaluate Chinese model families faster.
How does SmarToken billing work?
SmarToken uses prepaid USD wallet credit. Each API call records input tokens, output tokens, estimated cost and provider latency, then deducts usage from the account balance.