Moonshot AI
Kimi K3
Moonshot AI's open vision-language model with a 1M-token context, ready to serve across a multi-node Runpod Cluster.
Kimi K3
Kimi K2.6 & Kimi K2.7 Code: $4.00 per 1m output tokens, $0.95 per 1m input tokens
Kimi K3: $15.00 per 1m tokens, $3.00 per 1m input tokens
Kimi K3.1 Pro: $10.00 per 1m output tokens, $2.00 per 1m input tokens
Kimi K3.1 Pro Highspeed: $20.00 per 1m output tokens, $4.00 per 1m input tokens
Specifications
Provider
Moonshot AI
Modality
Vision + language
Task
Text generation
Parameters
Active parameters
Context
1M tokens
Precision
Quantized
Weights
2837
License
kimi-k3
Released
2026-06-12
Endpoint ID
moonshot-kimi
Weights on Hugging Face
moonshotai/Kimi-K3
Frequently asked questions
How do I use Kimi K3 on Runpod?
Call Kimi K3 as a Runpod Public Endpoint with your API key. Runpod hosts and scales the model, and you pay per use. You can also deploy the open weights on your own Serverless endpoint.
What GPU do I need to run Kimi K3?
Kimi K3 needs about 3,405 GB of GPU memory, more than one 8-GPU node, so it runs on a multi-node Runpod Cluster.
How much does Kimi K3 cost on Runpod?
Public Endpoints bill per use at the rates on this page. Serverless endpoints and Pods bill per second for the GPU you choose, with no charge while a Serverless endpoint is scaled to zero.
moonshotai-kimi-k3
moonshotai/Kimi-K3
moonshot-kimi
kimi-k3
Vision + language
2026-06-12 17:00
Multi-node cluster