Models

Kimi-K3: GPU memory and hosting cost

Open weights · 2779.93 billion parameters · Mixture of experts · 1,048,576-token context · licence other

No cloud sells this model per token: to use it you run it yourself, on GPU instances you rent.

Run it yourself

It needs 6949.8 GB of GPU memory at fp16, the precision it ships in.

More from Moonshot AI

• Kimi-K2.6
• Kimi-K2-Instruct
• Kimi-Linear-48B-A3B-Instruct
• Kimi-Linear-48B-A3B-Base
• Kimi-K2.7-Code
• Kimi-K2-Instruct-0905

Similar size, open weights

• DeepSeek-V4-Pro
• DeepSeek-V4-Pro-0813
• Qwen3.8-2.4T-A95B
• Qwen3.8-2.4T-A95B-FP8

List prices collected 5 October 2026, refreshed every three hours. No negotiated rates or taxes.