aitoken-hub

Posted on Jul 3

I Built a Free API Playground with 95+ LLM Models (No Signup Wall)

#ai #opensource #llm #api

The Problem I Was Tired Of Solving

I'm a solo developer who works with LLMs daily. Every week I'd discover another amazing Chinese model — Qwen, DeepSeek, GLM, Kimi, Yi — each one excelling at something different. But here's the pain: every provider has its own API format, its own SDK, its own quirks.

Switching between them felt like constantly changing keyboards.

I kept thinking: these models are genuinely great and surprisingly cheap, but they're fragmented. A developer shouldn't need to integrate 15 different APIs just to test which model works best for their use case.

So I built AI Token Hub — a single, unified API gateway that gives you OpenAI-compatible access to 95+ LLM models from China and beyond.

What Is AI Token Hub?

It's an OpenAI-compatible API proxy that routes requests to 95+ models behind a single endpoint. You keep using the OpenAI SDK you already know — just change the base_url and api_key, and suddenly you have access to models from:

Alibaba (Qwen series)
DeepSeek (DeepSeek-V3, DeepSeek-R1)
Zhipu AI (GLM-4 series)
Moonshot (Kimi)
01.AI (Yi series)
Baichuan
StepFun (Step series)
MiniMax
And many more...

The best part? No signup wall. You can test everything in the browser playground right now.

Models at a Glance

Here are some of the most popular ones available today:

Model	Provider	Context	Highlights
deepseek-v3	DeepSeek	64K	Top-tier reasoning, great for code
deepseek-r1	DeepSeek	64K	Advanced chain-of-thought
qwen-max	Alibaba	32K	Powerful general-purpose
qwen-plus	Alibaba	128K	Great balance of cost/performance
glm-4-plus	Zhipu AI	128K	Strong Chinese + English
moonshot-v1-128k	Moonshot	128K	Long-context specialist
yi-lightning	01.AI	16K	Fast and efficient
baichuan-4	Baichuan	32K	Chinese-optimized
step-2-16k	StepFun	16K	Versatile workhorse
abab6.5s-chat	MiniMax	245K	Massive context window
gemma-2-9b-it	Google	8K	Lightweight, great for edge
llama-3.3-70b-instruct	Meta	128K	Open-source powerhouse
doubao-1.5-pro-32k	ByteDance	32K	ByteDance's flagship
ernie-4.0-turbo-8k	Baidu	8K	Strong Chinese NLP
hunyuan-pro	Tencent	32K	Tencent's best

Full model list: https://aitoken-hub.github.io/aitoken-hub/

Show Me the Code

Python (using OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.example.com/v1",  # AI Token Hub endpoint
    api_key="your-api-key"
)

response = client.chat.completions.create(
    model="deepseek-v3",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Explain quantum computing in 3 sentences."}
    ],
    temperature=0.7
)

print(response.choices[0].message.content)

cURL

curl -X POST "https://api.example.com/v1/chat/completions" \
  -H "Authorization: Bearer your-api-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-max",
    "messages": [
      {"role": "user", "content": "Write a haiku about programming"}
    ],
    "temperature": 0.8
  }'

That's it. No new SDK. No new documentation to learn. Just swap the base URL.

Try It Right Now — No Signup

I built a Playground so you can test any model in seconds, no account needed:

👉 Open the Playground

Pick a model, type a prompt, see results. Compare outputs side by side. Find the best model for YOUR use case.

Why Chinese Models?

Here's what most Western developers don't realize:

Price. Models like DeepSeek-V3 cost a fraction of GPT-4o per token. Qwen-Plus offers 128K context at prices that would make OpenAI blush.
Quality. Qwen and DeepSeek consistently rank near the top of open benchmarks. They're not "budget alternatives" anymore.
Specialization. Many of these models are exceptionally good at tasks involving Asian languages, coding, math reasoning, and long-context understanding.

The barrier was never quality — it was accessibility. That's the gap AI Token Hub closes.

What's Next

More models being added weekly
Streaming support across all endpoints
Better playground UX (prompt templates, comparison mode)
Community-contributed model evaluations

DEV Community