Pricing
Pay for usage.
Save through efficiency.
No minimums, no hidden fees. The free tier is enough to prototype.
20% cheaper than direct cloud, available on signup.
- 1M token free credit
- OpenAI-compatible API
- All leading models
- Community support
50M tokens / month — save up to 50%.
- 50M tokens / month
- Semantic cache acceleration
- Ticket support (24h)
- VAT general invoice
Dedicated SLA, self-hosting and compliance delivery.
- Dedicated routing policy
- SLA 99.9%
- Private licensing
- MLPS compliance advisory
- VAT special invoice
Per-model pricing
Token pricing details
| Model | Input / 1K | Output / 1K | Cache-hit price |
|---|---|---|---|
| Qwen3-Max | ¥0.012 | ¥0.036 | ¥0.001 |
| Qwen3-Turbo | ¥0.002 | ¥0.006 | ¥0.0002 |
| DeepSeek-V3 | ¥0.004 | ¥0.012 | ¥0.0004 |
| DeepSeek-R1 | ¥0.016 | ¥0.064 | ¥0.002 |
| Hunyuan-Pro | ¥0.024 | ¥0.072 | ¥0.003 |
| Doubao-Pro | ¥0.0008 | ¥0.002 | ¥0.0001 |
| BGE-M3 (embed) | ¥0.0005 | — | — |
* Cache-hit pricing applies when the semantic cache returns the response — up to 90% savings.
Plan comparison
Feature overview
| Feature | PAYG | Monthly | Enterprise |
|---|---|---|---|
| API calls | Unlimited | Unlimited | Unlimited |
| Available models | All leading models | All leading models | All + private models |
| Concurrency | 20 QPS | 100 QPS | Custom |
| Semantic cache | ✓ | ✓ | ✓ |
| Smart routing | ✓ | ✓ | ✓ Custom policy |
| Batch inference | ✓ | ✓ | ✓ |
| Content moderation | ✓ | ✓ | ✓ |
| Usage monitoring | Basic dashboard | Advanced dashboard | Enterprise reports + SLA |
| Support | Community | Tickets (24h) | Dedicated CSM + 24×7 |
| Invoicing | — | VAT general | VAT special |
| Self-hosting | — | — | ✓ |
| MLPS advisory | — | — | ✓ |
Billing details
Billing rules
Pay-as-you-go
Billed per actual input/output tokens, accurate to 1K tokens. No minimums; unused credit never expires.
Monthly billing
Fixed monthly fee covers 50M tokens. Overage is billed at 80% of PAYG. Credit resets at month-end; auto-renewed.
Cache pricing
Semantic-cache hits bill at the cache-hit rate — 1/10 to 1/20 of the standard price. No impact on output quality.
Billing cycle
Settled by calendar month. Invoices generated on the 1st. Alipay, corporate transfer, and credit lines (enterprise) supported.
FAQ
About pricing
Why are you cheaper than direct cloud?+
Semantic caching, batching and multi-cloud price arbitrage cut our unit cost, and we pass part of the savings to customers.
How do I get the free credit?+
1M tokens are granted automatically after KYC, usable across all models, valid for 30 days.
Do you provide invoices?+
VAT general invoices from the Monthly plan; VAT special invoices on Enterprise. Issued within 7 business days of payment.
How do I control costs?+
Budget alerts and automatic throttling can be configured per API key, tenant or model.
Do you support self-hosting?+
Enterprise plans deliver a source-licensed install into your network, with Xinchuang hardware and GM-crypto support.
Do unused monthly tokens roll over?+
Monthly credit is per calendar month; unused tokens reset at month-end. Pick the plan that matches your usage.
How is Enterprise priced?+
Enterprise pricing is custom-quoted based on volume, deployment scale and SLA. Contact sales for a proposal.
How do I upgrade or downgrade?+
Change plans anytime in the console. Monthly upgrades are immediate; downgrades apply next month. PAYG needs no change.
Still deciding?
The free tier is enough to evaluate every leading model.