Guide · Updated November 2026

Use Qwen models via API with a credit card — no Chinese platform required

Alibaba's Qwen family is one of the strongest open-weight model lineups anywhere — and one of the hardest to actually pay for from outside China. NovaToken gives you a single OpenAI-compatible endpoint for Qwen (plus DeepSeek, GLM and Kimi), billed in USD on any card.

The short version: register with an email → buy $5 of credits with any card → set base_url to https://api.novatoken.dev/v1. Streaming, tool calling and JSON mode all work out of the box.

Why Qwen, and why it's hard to get

Qwen models punch far above their price: the 7B-class instruction model handles chat, classification and tool calling reliably, with roughly a two-second first token and no thinking pause. Developers use them for chatbots, data extraction pipelines, translation, and as cheap fallback models behind a bigger frontier model.

The catch is access. The usual routes — Alibaba Cloud international, or Chinese inference platforms — mean Chinese accounts, Alipay/WeChat top-ups, or enterprise verification. For a solo developer with a Visa card, that's a dead end. NovaToken exists to fix exactly that:

  1. No regional gatekeeping — email registration only, works worldwide.
  2. Card payments in USD — $5 / $20 / $50 prepaid packs, credits never expire.
  3. Zero code changes — if your code talks to OpenAI, it already talks to us.

Start calling Qwen in one minute

# curl
curl https://api.novatoken.dev/v1/chat/completions \
  -H "Authorization: Bearer sk-your-novatoken-key" \
  -H "Content-Type: application/json" \
  -d '{"model":"Qwen/Qwen2.5-7B-Instruct","messages":[{"role":"user","content":"Hello!"}]}'
# Python (openai SDK)
client = OpenAI(api_key="sk-your-novatoken-key", base_url="https://api.novatoken.dev/v1")

resp = client.chat.completions.create(
    model="Qwen/Qwen2.5-7B-Instruct",
    messages=[{"role": "user", "content": "Summarise this email in two sentences: ..."}],
)
# first token in ~2s, no thinking pause

Full model line-up

ModelBest forPrice / 1M tokens
Qwen2.5-7B-InstructFast chat, classification, tool calling$0.10 in / $0.30 out
DeepSeek-V4-FlashReasoning, long documents (1M context)$0.30 in / $1.20 out
GLM-5.3-FlashBilingual tasks, agents$0.15 in / $0.50 out
Kimi-K2.7-CodeCode, agentic workflows (262K context)$1.20 in / $5.00 out

Try it for $5

Fifty million input tokens on Qwen2.5-7B-Instruct for five dollars — enough to prototype a full product. No subscription, no expiry, top up again whenever you need.

Buy $5 of credits Create free account

FAQ

Which Qwen version do you serve?

Currently Qwen2.5-7B-Instruct (fast, no thinking mode). Larger Qwen models are added as demand grows — email [email protected] if you need a specific one.

Is it compatible with the OpenAI SDK?

Yes. Chat completions, streaming (SSE), function/tool calling and JSON mode are supported in every language with an OpenAI-compatible client.

How fast is it?

Qwen2.5-7B-Instruct answers with roughly a 2-second first token. The reasoning models (DeepSeek, GLM, Kimi) pause a few seconds while thinking; with streaming the first bytes still arrive fast.

Do you store my prompts?

No. Billing metadata only, purged on a rolling basis. Request bodies are not retained.

What payment methods work?

Major credit and debit cards worldwide, charged in USD. Prepaid packs of $5 / $20 / $50.