Latest Content
Olmo 3.1 32B Think icon

Olmo 3.1 32B Think

NVIDIA
Olmo 3.1 32B Think is a dense transformer language model and a successor to Olmo 3, designed for advanced reasoning, mathematics, and complex multi-step problem solving. It is built on a 32B parameter architecture with 64 layers, 5,120 hidden size, and 40 attention heads, using grouped query attention with 8 key-value heads for efficient inference. The model supports up to a 65K context window with YaRN-based RoPE scaling and combines sliding window attention with a 4K window and periodic full attention layers for improved long-context reasoning. Trained with extended reinforcement learning runs, it delivers strong performance on deep reasoning, planning, and iterative problem-solving tasks.
TypeDense LLM
CapabilitiesText Generation, Instruction Following, Reasoning, Mathematical Reasoning+3 more
Release Date15 December, 2025
Links
Paper|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:v0.18.1 
 allenai/Olmo-3.1-32B-Think 
  --tensor-parallel-size 8 
 --max-model-len auto 
  --max-num-batched-tokens 65536 
 --gpu-memory-utilization 0.95 
 --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