Official upstream keys•Automatic failover•Nothing stored•Works with Cursor / Claude Code / Cline•No prompt inspection•Pay per token•One key, every model•
OpenAI Compatible · /v1 Ready

Token API: one key
for every major model

Open models like DeepSeek-V4.1-Flash-1017 and GLM-5.3, on official upstream keys — no audit logging, billed per token, less hassle than going direct.

Get your passport
Base URLhttps://hcxai.dpdns.org/v1

Model passport

Model Passport · /v1 Access

Lines healthy (99.99%)

Request route

4 models ready

Your app / IDE

Cursor, Claude Code, scripts, services

client

Token API gateway

Unified auth, smart routing, instant failover

/v1

DeepSeek-V4.1-Flash-1017

Current latency 15ms · strong reasoning

ready

Live uptime 99.99%

PASS-2026-V1

Admit one
No. 0001

Why This Gateway

Why route everything through one gateway

One SDK, one key, one bill.

ModelsMODEL

One key, every provider

Qwen, DeepSeek, GLM, Kimi and more behind a single OpenAI-compatible endpoint, all on official upstream APIs. Switching models is a one-field change.

ReliabilitySLA

Automatic failover

Several upstreams stay online at once. When one wobbles, traffic moves to a backup line and your app never notices — no retry logic to write.

Protocol/V1

Drop-in OpenAI

Existing SDKs, IDE plugins and scripts keep working unchanged. Point the base URL here and leave every other parameter alone.

PrivacySAFE

Official keys, nothing retained

Every upstream is a paid official API. Requests are not inspected in the path, and neither request nor response bodies are written to storage.

Live line status

Qwen3.8-Flash12ms

Qwen3.8-Flash

Alibaba · balanced, low latency

Uptime 99.99%

DeepSeek-V4.1-Flash-101715ms

DeepSeek-V4.1-Flash-1017

DeepSeek · strong reasoning

Uptime 99.99%

Qwen3.7-Plus18ms

Qwen3.7-Plus

Alibaba · long context

Uptime 99.98%

GLM-5.221ms

GLM-5.2

Zhipu · cost efficient

Uptime 99.98%

GLM-5.324ms

GLM-5.3

Zhipu · general purpose

Uptime 99.97%

Kimi-K2.7-Code27ms

Kimi-K2.7-Code

Moonshot · long context code

Uptime 99.96%

1-minute setup

If you can use OpenAI,
you can use Token API.

No business logic to rewrite: swap the base URL for this site and use a model id that is actually on sale.

Base URLhttps://hcxai.dpdns.org/v1
Get an API key
quick-start.py
1from openai import OpenAI
2
3client = OpenAI(
4 base_url="https://hcxai.dpdns.org/v1",
5 api_key="sk-your-key",
6)
7
8stream = client.chat.completions.create(
9 model="deepseek-v4.1-flash-1017",
10 messages=[{"role": "user", "content": "Explain what an API gateway does in one sentence"}],
11 stream=True,
12)
13
14for chunk in stream:
15 delta = chunk.choices[0].delta.content
16 if delta:
17 print(delta, end="", flush=True)
// Press the button on the left — no key needed to try

Gateway · Auto Routing

Model Passport for Vibe Coding

Three steps to connect

From sign-up to your first request

Already have a key? Skip to step three — the base URL is right on the home page.

01

Create an API key

After signing up, issue one yourself under Tokens in the console. Revoke it any time.

02

Swap the base URL

Point your client's base_url at this site's /v1 and leave every other parameter alone.

03

Send a request

Your existing OpenAI SDK code works as-is — the first call already returns a result.

Get your passport now

Usable right after sign-up. Pay for what you use, with no starter package to buy — and nothing you send is retained.

Sign up free