Model and agent development
Token API
Designed for large language model applications, our standardized service uses tokens as a unified unit of measurement and delivers reliable API access for LLM inference, conversational AI, and content generation. It is compatible with mainstream API specifications and supports on-demand access and pay-as-you-go billing. Cross-border connectivity is also available for international token service delivery.
The platform brings together leading model families, including DeepSeek, Qwen, Doubao, Kimi, and more.
| Series | Model Name | Type | Unit | Price (HKD) |
|---|---|---|---|---|
| DeepSeek Series | DeepSeek-V3 | Input | Per million tokens | 1.344 |
| Output | Per million tokens | 5.376 | ||
| Doubao Series | Doubao 1.5 Lite 32K | Input (cache hit) | Per million tokens | 0.0600 |
| Input | Per million tokens | 0.300 | ||
| Output | Per million tokens | 0.600 | ||
| Kimi Series | Kimi K2.6 Cache | Input (cache hit) | Per million tokens | 1.100 |
| Input | Per million tokens | 6.500 | ||
| Output | Per million tokens | 27.00 |
Built to move your work forward
- Open model catalog
- Token-level billing
- Low-latency endpoints