hy4-preview
Tencent · Priced per token
The Hybrid-4 (Hy4) preview version boasts 770B total parameters and 49B activated parameters, with targeted optimizations for Agent, coding and productivity scenarios to enhance capabilities in complex task understanding, planning, tool calling and continuous execution. Compared with the previous-generation model, Hy4 preview delivers further improvements in multi-step Agent operations, code development and productivity-related tasks, featuring better performance in task decomposition, context coherence, instruction following and long-chain execution. In coding scenarios, it has further strengthened capabilities in code understanding, generation, modification and complex engineering task processing; in productivity scenarios, it focuses on upgrading capabilities including document processing, information analysis, office automation, game development, web page generation and cross-tool collaboration. Hy4 preview is applicable to coding Agents, complex tool calling, and various Agent workflows that require multi-step planning and continuous execution, offering more reliable task completion capabilities for complex tasks in real business scenarios.
Pricing
Base price
- Input: $0.8946 per 1M tokens
- Output: $2.6837 per 1M tokens
- Cached input: $0.4478 per 1M tokens
List rate at a 1x group multiplier, before any account-specific discount. The effective rate is this figure times the multiplier of the group the account is in.
Pricing by group
| Group | Multiplier | Input | Output |
|---|---|---|---|
| default | 1x | $0.8946 | $2.6837 |
API access
- openai — POST /v1/chat/completions
Available to groups: default