15+ Chinese AI Models · Enterprise Infrastructure · Stable

One KeyAccess All Chinese LLMs

Helin API is OpenAI-compatible, unifying DeepSeek, Qwen, GLM, Kimi, iFlytek Spark, MiniMax and more into a single platform. Manage tokens, billing, logs and pricing in one console.

Sign Up - Get Free CreditsAPI Docs
Base URLhttps://helinapi.com
/v1/chat/completions·/v1/models·/v1/embeddings·/v1/images/generations·/v1/audio/transcriptions·/v1/chat/completions·/v1/models·/v1/embeddings·/v1/images/generations·/v1/audio/transcriptions
<50ms
IEPL Latency
99.7%
30-Day Uptime
15+
Chinese LLMs
Zero
Data Retention
Model Marketplace

DeepSeek Full Lineup

OpenAI-compatible. Pay per token. More vendors coming soon.

Flagship
DeepSeek
V4-Pro
Latest flagship model with 1M context window. Superior reasoning, comprehension and generation. Surpasses GPT-5.5 at 1/10 the price.
1M ContextThinking Mode
¥4.50
/ 1M input tokens · ¥9.00 output
Popular
DeepSeek
V4-Flash
Best value model. 1M context. Integrated R1 reasoning at 33% lower cost than V3.2.
1M ContextThinking Mode
¥1.50
/ 1M input tokens · ¥3.00 output
HOT
DeepSeek
V3.2
Proven general-purpose model, 160K context
160K
¥3.00
/ 1M input · ¥4.50 output
DeepSeek
R1
Deep reasoning specialist, 32K chain-of-thought
128KReasoning
¥6.00
/ 1M input · ¥24.00 output
Value
DeepSeek
R1-Distill-Qwen-32B
Distilled R1, 32B params, beats OpenAI-o1-mini
32B
¥0.03
/ request
DeepSeek
R1-Distill-Qwen-14B
Distilled R1, 14B, balanced performance
14B
¥0.0225
/ request
DeepSeek
R1-Distill-Qwen-7B
Distilled R1, 7B, lightweight reasoning
7B
¥0.015
/ request
DeepSeek
R1-Distill-Qwen-1.5B
Distilled R1, 1.5B, ultra-lightweight
1.5B
¥0.015
/ request
DeepSeek
Coder-33B
Code specialist, 87% code training data, 16K window
16KCode
¥0.03
/ request
DeepSeek
R1-Search
R1 with web search, real-time info retrieval
Web
¥9.00
/ 1M tokens
DeepSeek
V3-Search
V3 with web search
Web
¥4.50
/ 1M tokens
DeepSeek
OCR-2
High-precision image text recognition
OCR
¥0.03
/ request

Qwen · GLM · Kimi · iFlytek Spark · MiniMax coming soon.

Platform

Enterprise AI Gateway

Beyond API relay - a complete AI model access solution with infrastructure, compliance and developer tooling.

⚡

IEPL Accelerated

Dedicated line to upstream providers, <50ms latency. Auto failover and load balancing for high concurrency.

🛡

Full Compliance

ICP licensed, Level-3 certified. Chinese models only. Zero data retention - prompts and completions are never stored.

🔧

Drop-in Replacement

Swap Base URL and API Key. Chat, Stream, Function Calling, Vision all supported. Works with all OpenAI-compatible tools.

Pricing

Pay As You Go

Real-time token billing. Balance never expires. Enterprise invoicing and VAT fapiao available.

Starter
¥50
Instant access
  • All models
  • Real-time billing
  • No expiry
Get Started
Popular
¥500
+ ¥50 bonus
  • All models
  • Priority support
  • Bank transfer
  • VAT invoice
Get Started
Business
¥1,000
+ ¥150 bonus
  • All models
  • Dedicated manager
  • Custom pricing
  • Concurrency SLA
Contact Us
Enterprise
¥5,000
+ ¥1,000 bonus
  • All models
  • 99.9% SLA
  • Private deployment
  • 24/7 support
Contact Us
FAQ

Questions?

Which models are supported?
DeepSeek V4-Pro/V4-Flash/V3.2/R1, R1-Distill series, Coder-33B, OCR-2 and more. 15+ Chinese LLMs with more being added.
How do I integrate?
Set Base URL to https://helinapi.com and use your API Key from the dashboard. Compatible with OpenAI SDK, Cherry Studio, LobeChat, Cursor, Claude Code and all OpenAI-format tools.
Is my data secure?
Prompts and completions are never stored. Licensed operator with Level-3 security certification. Chinese models only - no cross-border data flows.
Do you support enterprise purchasing?
Yes. Bank transfer, VAT invoices, SLA agreements available. Contact support to set up an enterprise account.

Ready to Start?

Sign up for free credits. No credit card required. Two minutes to integrate.

Sign Up FreeRead the Docs