Latest Content
Qwen3.5 4B icon

Qwen3.5 4B

NVIDIA
Qwen3.5-4B is a multimodal causal language model built with a 4B parameter architecture. It uses a 32-layer transformer with 16 attention heads, 4 KV heads, and a 2,560 hidden size, paired with a 9,216 intermediate dimension. The model supports a native 262K token context window, extendable beyond 1M tokens, and integrates a 24-layer vision encoder with 1,024 hidden size. It combines Gated DeltaNet and attention layers, enabling efficient long-context reasoning, strong multimodal understanding, and broad multilingual capability across 200+ languages.
TypeVision-Language Model
CapabilitiesText Generation, Instruction Following, Reasoning, Mathematical Reasoning+6 more
Paper/Blog
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 --gpus all 
 --shm-size 128g 
 -p 8000:8000 
 -v ~/.cache/huggingface:/root/.cache/huggingface 
 -e HF_TOKEN='YOUR_HF_TOKEN' 
 --ipc=host 
 lmsysorg/sglang:v0.5.9 
 python3 -m sglang.launch_server 
 --model-path Qwen/Qwen3.5-4B 
 --host 0.0.0.0 
 --port 8000 
 --max-prefill-tokens 65536 
 --max-running-requests 1024 
 --tp 8 
 --tool-call-parser qwen3_coder 
 --reasoning-parser qwen3 
 --mem-fraction-static 0.95 
 --trust-remote-code
Note

Model is served in multimodal mode by default; to restrict the engine to text-only processing, use the --language-only flag.

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