TokenRelay
DeepSeek V3 from $0.08/1M tokens · 97% cheaper than GPT-4o

AI APIs at
Chinese Prices

Access DeepSeek, Qwen, GLM and 30+ models at a fraction of Western API costs. Fully OpenAI-compatible — change one line, save 90%.

30+
Models Available
$0.08
DeepSeek V3 /1M input
$0.01
Qwen Flash /1M input
90%
Savings vs OpenAI

Why TokenRelay?

Chinese LLM APIs at unbeatable prices — without the Great Firewall headache

90% Cheaper

DeepSeek V3 at $0.08/1M vs GPT-4o at $2.50/1M. Qwen Flash at $0.01/1M. Same quality, fraction of the cost.

30+ Top Models

DeepSeek V3/R1, Qwen 3.x, GLM-4.7, Kimi K2, Stepfun — China's best AI models, all in one API.

No Markup, No Middlemen

We source directly from Chinese providers. No Western SaaS tax. You pay what the model costs, plus a small fee.

Global Payments

Pay with USDT (TRC20) or international credit cards via Stripe. Alipay/WeChat also available.

OpenAI Compatible

100% OpenAI API format. Swap your base URL and API key — no code changes needed. Works with all existing SDKs.

Fast & Reliable

US-based VPS with 30Mbps bandwidth. Smart routing auto-selects the fastest upstream. 99.9% uptime, real-time monitoring.

Compare & Save

Chinese models deliver comparable quality at a fraction of the price

TokenRelay
$0.01
/1M input tokens
vs
OpenAI GPT-4o
$2.50
/1M input tokens
deepseek/deepseek-chat SAVE 99.6%
DeepSeek
qwen/qwen3.6-flash
Alibaba Cloud · $0.003/1M vs GPT-4o mini $0.15
z-ai/glm-4.7-flash
Zhipu AI · $0.003/1M — best price/performance
deepseek/deepseek-r1
DeepSeek · $0.02/1M reasoning — rivaling o1 at $15
moonshotai/kimi-k2
Moonshot AI · cutting-edge Chinese model
stepfun/step-3.7-flash
Stepfun · fast & budget-friendly
View Full Pricing →

Ready to Save 90% on AI APIs?

Sign up for free trial credit and start using DeepSeek, Qwen & more today

Get Started Free