Llama Nemotron Ultra 253BOpen Weight
Released
Llama Nemotron Ultra 253B is an AI model released by NVIDIA on Monday, Apr 7 2025, 20 days after Llama Nemotron (Nano/Super). It is an open-weight model — the trained weights are available to download and run. It is a 253B parameterThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. model with a 128k token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover LiveCodeBench, and GPQA Diamond.
Get NVIDIA releases and AI news.
Every other Monday
Benchmarks
Coding
Reasoning & science
Compare Llama Nemotron Ultra 253B with
Suggested comparisons
Llama Nemotron Ultra 253BvsNemotron 3.5 Bolt E4B 4GBLlama Nemotron Ultra 253BvsGPT-6.1 TerraLlama Nemotron Ultra 253BvsClaude Fable 5.2Llama Nemotron Ultra 253BvsGemini 4 CarbonLlama Nemotron Ultra 253BvsMuse EmberLlama Nemotron Ultra 253BvsGrok 4.7Llama Nemotron Ultra 253BvsDeepSeek-V4.1-Flash-1017Llama Nemotron Ultra 253BvsMistral Large 4Llama Nemotron Ultra 253BvsKimi K3.1 ProLlama Nemotron Ultra 253BvsGLM-5.3-FlashLlama Nemotron Ultra 253BvsQwen3.8-Max-0902About
Llama Nemotron Ultra 253B was released in April 2025 as the largest tier of NVIDIA's Llama Nemotron reasoning family. It was derived from Meta's Llama 3.1 405B using neural architecture search to reduce the model to approximately 253 billion parameters. The aim was to retain strong capability while making deployment more efficient than the original foundation model.
The model supported reasoning and non-reasoning modes and a 128,000-token context. NVIDIA documented deployment on a single eight-GPU H100 node, with requirements dependent on precision and serving configuration. The released weights provided a large self-hosted reasoning model derived from Llama, rather than a consumer assistant service.
Frequently asked questions
Llama Nemotron Ultra 253B was released by NVIDIA on Monday, Apr 7 2025.