Kimi K3 has 2.8 trillion parameters. It is built on the Kimi hybrid linear attention mechanism (Kimi Delta Attention) and Attention Residuals technology. With native support for visual understanding and a context window of up to 1 million tokens, it is designed for cutting-edge intelligent scenarios such as software engineering, knowledge work, and deep reasoning.
Différents groupes de tokens ont différents prix. Unité : million de tokens (M)
| Groupe de tokens | Description | Ratio du groupe | Prix d'entrée | Prix de sortie | Lecture du cache |
|---|---|---|---|---|---|
| sale | 特价模型分组 | 0.5x | $10/M | $50/M | $1/M |