Latest Content
Mistral Small 4 119B 2603 icon

Mistral Small 4 119B 2603

NVIDIA
Mistral Small 4 119B 2603 is a hybrid multimodal Mixture-of-Experts model that unifies instruction-following, reasoning, and coding capabilities into a single system. It contains 119B parameters with ~6.5B active per token, using 128 experts with 4 selected per token across a 36-layer architecture with 4,096 hidden size and 32 attention heads. Supporting up to a 256K context window, it enables dynamic reasoning modes, vision understanding, and strong agentic tool calling with structured outputs, making it well-suited for high-throughput, real-world AI systems.
TypeVision-Language Model
CapabilitiesText Generation, Instruction Following, Reasoning, Mathematical Reasoning+6 more
Release Date16 March, 2026
Links
Blog|HF Model Card
LicenseApache 2.0

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 
 mistralai/Mistral-Small-4-119B-2603 
  --tensor-parallel-size 8 
 --max-model-len auto 
  --max-num-batched-tokens 32768 
 --gpu-memory-utilization 0.90 
 --tool-call-parser mistral 
 --reasoning-parser mistral 
 --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