Latest Content
Kimi K2.6 icon

Kimi K2.6

NVIDIA
Kimi K2.6 is a native multimodal Mixture-of-Experts model designed for long-horizon coding, agentic workflows, and large-scale autonomous task orchestration, succeeding Kimi K2.5. It features 1T total parameters with ~32B active, 61 layers, 7168 hidden size, and 64 attention heads, activating 8 of 384 experts plus 1 shared expert per token. The model supports a 256K context window with MLA attention and YARN RoPE scaling. It integrates a 400M-parameter MoonViT vision encoder and enables large-scale agent swarms for parallel task execution.
TypeVision-Language Model
CapabilitiesText Generation, Instruction Following, Reasoning, Mathematical Reasoning+6 more
Release Date20 April, 2026
Links
Blog|HF Model Card
LicenseModified MIT

Inference Instructions

Deploy and run this model on NVIDIA B200 GPUs using the command below. Copy the command to get started with inference.

CONSOLE
docker run -it --rm 
 --runtime=nvidia 
 --gpus all 
 --ipc=host 
 --shm-size=128g 
 -p 8000:8000 
 -v ~/.cache/huggingface:/root/.cache/huggingface 
 -e HF_TOKEN='YOUR_HF_TOKEN' 
 vllm/vllm-openai:v0.19.1 
 moonshotai/Kimi-K2.6 
  --tensor-parallel-size 8 
 --mm-encoder-tp-mode data 
 --max-model-len auto 
  --gpu-memory-utilization 0.90 
 --tool-call-parser kimi_k2 
 --reasoning-parser kimi_k2 
 --enable-auto-tool-choice 
 --max-num-seqs 1024 
 --trust-remote-code

Model Benchmarks

Each model was tested with a fixed input size and total token volume while increasing concurrency to measure serving performance under load.

ITL vs Concurrency

Time to First Token

Throughput Scaling

Total Tokens/sec vs Avg TTFT

Vultr Cloud GPU

NVIDIA HGX B200

Deploy NVIDIA B200 on Vultr Cloud GPU

Deploy NVIDIA B200 instancesContact Sales
How to Deploy Kimi K2.6 on NVIDIA GPUs | Vultr Docs