Latest Content
Mistral Large 3 675B Instruct 2512 icon

Mistral Large 3 675B Instruct 2512

NVIDIA
Mistral-Large-3-675B-Instruct-2512 is a large-scale multimodal Mixture-of-Experts (MoE) model designed for instruction following, agentic workflows, and long-context understanding. It has 675B total parameters with about 41B activated, using a 61-layer transformer with 128 attention heads and a 7,168 hidden size. The architecture includes 128 experts with 4 selected per token and supports a context window of ~294K tokens. It also integrates a 2.5B vision encoder and FP8 quantization, enabling efficient deployment and strong performance across text and vision tasks.
TypeVision-Language Model
CapabilitiesText Generation, Instruction Following, Reasoning, Mathematical Reasoning+6 more
Group Release DateDecember 1, 2025
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:v0.18.0 
 mistralai/Mistral-Large-3-675B-Instruct-2512 
  --tensor-parallel-size 8 
 --max-model-len auto 
 --reasoning-parser mistral 
 --tool-call-parser mistral 
 --tokenizer_mode mistral 
 --config_format mistral 
 --load_format mistral 
 --enable-auto-tool-choice 
  --gpu-memory-utilization 0.9 
  --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