Kimi K3 on vLLM: Up to 370 Tokens/sec

Kimi K3 achieves up to 370 tokens per second on the vLLM inference framework, demonstrating high efficiency.

• vLLM
• Performance
• Kimi K3