MLPerf Benchmarks
The NVIDIA AI platform achieves world-class performance and versatility in both training and inference, enabled by extreme co-design.
What Is MLPerf?
MLPerf benchmarks are designed to provide unbiased evaluations of training and inference performance for hardware, software, and services. Developed by MLCommons, a consortium of AI leaders from academia, research labs, and industry, these evaluations are all conducted under prescribed conditions. To stay on the cutting edge of industry trends, MLPerf continues to evolve with new models, workloads, and measurement methods that reflect how AI is developed and deployed.
MLPerf Training
MLPerf Training measures the time to train models to a specified quality target. Its workloads span large language model pretraining, fine-tuning, and post-training, as well as text-to-image generation and recommendation.
MLPerf Endpoints
MLPerf Endpoints measures inference performance through live API endpoints across a wide variety of model architectures, including dense and mixture-of-expert (MoE) large language models, vision language models, generative recommenders, and more. The benchmark shows how throughput, responsiveness, and time to first token change as the number of concurrent users increases, helping organizations evaluate systems and services for their deployment needs. Results are published on an ongoing basis, keeping comparisons current as AI infrastructure evolves.
NVIDIA MLPerf Benchmark Results
-
Training
-
Inference
The NVIDIA platform excelled in delivering the fastest time to train and the highest performance per GPU in MLPerf Training v6 benchmarks. This round, NVIDIA submitted results on both GB200 NVL72 and GB300 NVL72 systems. At the same scale, GB300 NVL72 delivered up to 1.6x faster training than GB200 NVL72. This round added two new MoE pretraining workloads, DeepSeek-V3 671B and GPT-OSS-20B, and NVIDIA set performance records on both. On DeepSeek-V3 671B, NVIDIA scaled to 8,192 GPUs using GB200 NVL72 systems, the largest-scale NVIDIA Blackwell-based submission in MLPerf Training to date.
NVIDIA Blackwell Platform Raises the Bar for Performance and Scale
MLPerfTraining v5.0, v5.1, and v6.0 results retrieved from www.mlcommons.org on June 16, 2026. MLPerf GPU scale results from entries 5.0-0004, 5.1-004, 6.0-0001, 6.0-0005, and 6.0-0014. MLPerf Blackwell training comparison from the following entries: 6.0-0006, 6.0-0013, 6.0-0017, 6.0-0018, 6.0-0078, and 5.1-0072. The MLPerf name and logo are registered and unregistered trademarks of MLCommons Association in the United States and other countries. All rights reserved. Unauthorized use is strictly prohibited. See www.mlcommons.org for more information.
Annual Rhythm and Extreme Co-Design for Sustained Training Leadership
The NVIDIA platform delivered the fastest time to train on every MLPerf Training v6 benchmark, with innovations across chips, systems, and software enabling sustained training performance leadership, as shown on industry-standard, peer-reviewed performance data.
Max-Scale Performance
| Benchmark | Time to Train |
|---|---|
| DeepSeek-v3 671B | 2.02 minutes |
| GPT-OSS-20B | 7.43 minutes |
| Llama 3.1 405B | 7.07 minutes |
| Llama 2 70B LoRA | 0.40 minutes |
| Llama 3.1 8B | 4.46 minutes |
| FLUX.1 | 17.1 minutes |
| DLRM-dcnv2 | 0.67 minutes |
MLPerf™ Training v6.0 retrieved from www.mlcommons.org on June 16, 2026, from the following entries: 6.0-0005, 6.0-0102, 6.0-0001, 6.0-0102, 6.0-0015, 6.0-0101, and 6.0-0062. The MLPerf name and logo are trademarks of MLCommons Association in the United States and other countries. All rights reserved. Unauthorized use is strictly prohibited. See www.mlcommons.org for more information.
NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut
Three factors are central to AI inference economics: system performance, infrastructure scaling efficiency, and continuous software optimization. Higher performance means more tokens generated, driving higher revenue. Efficient scaling means throughput grows proportionally as hardware is added, requiring fewer resources to serve users. Continuous optimization means more value from infrastructure already deployed.
The NVIDIA platform is purpose-built to advance all three, and MLPerf Inference v6.1 results show how:
- Vera Rubin NVL72 debuts with leading performance: In its first MLPerf Inference submission, Vera Rubin NVL72 delivers up to 3.7x higher throughput than GB300 NVL72.
- GB300 NVL72 scales with leading efficiency: A 288-GPU submission across four GB300 NVL72 racks achieved 99% scaling efficiency, with throughput growing nearly linearly from a single-rack baseline.
- Software optimizations drive continuous gains: v6.1 software advances delivered up to 1.6x higher performance over v6.0, with further gains from post-submission optimization.
Performance, scaling efficiency, and software velocity together determine long-term AI infrastructure economics.
MLPerf Inference v6.1, Closed Division. Results retrieved from www.mlcommons.org on Sept. 16, 2026. NVIDIA platform results from the following entries: 6.1-0106, 6.1-0073, and 6.1-0074. The MLPerf name and logo are registered and unregistered trademarks of MLCommons Association in the United States and other countries. All rights reserved. Unauthorized use is strictly prohibited. See www.mlcommons.org for more information.
The Technology Behind the Results
The complexity of AI demands a tight integration between all aspects of the platform. As demonstrated in MLPerf’s benchmarks, the NVIDIA AI platform delivers leadership performance with the world’s most advanced GPU, powerful and scalable interconnect technologies, and cutting-edge software—an end-to-end solution that can be deployed in the data center, in the cloud, or at the edge with amazing results.
Optimized Software That Accelerates AI Workflows
NVIDIA Dynamo is an open source distributed inference-serving framework to deploy models in multi-node environments at AI-factory-scale. It streamlines distributed serving by disaggregating inference, optimizing routing, and extending memory through data caching to cost-effective storage tiers.
Dynamo works by disaggregating (separating) the prefill and decoding phases of LLM inference across different GPUs, allowing for independent optimization and higher throughput. It was featured prominently in the MLPerf Inference v5.1 benchmarks, demonstrating superior performance in Llama 3.1 405B Interactive and DeepSeek-R1 reasoning tests.
Leadership-Class AI Infrastructure
Achieving world-leading results across training and inference requires infrastructure that’s purpose-built for the world’s most complex AI challenges. The NVIDIA AI platform delivered leading performance powered by the NVIDIA Blackwell and Blackwell Ultra platforms, including the NVIDIA GB300 NVL72 and GB200 NVL72 systems, NVLink and NVLink Switch, NVIDIA Quantum InfiniBand and NVIDIA Spectrum-X Ethernet scale-out networking. These are at the heart of AI factories powered by the NVIDIA data center platform, the engine behind our benchmark performance.
In addition, NVIDIA DGX™ systems offer the scalability, rapid deployment, and incredible compute power that enable every enterprise to build leadership-class AI infrastructure.
Learn more about our data center training and inference performance.
Reasoning LLMs
MLPerf Inference uses:
DeepSeek-R1 with samples sourced from the AIME, MATH500, GPQA Diamond, MMLU-Pro, and LiveCodeBench datasets.
GPT-OSS-120B with samples from the AIME 2024, LivecodeBench v6, and GPQA Diamond datasets.
Vision Language Model
MLPerf Inference uses the Qwen3-VL-235B-A22B-Instruct model with the Shopify Product Catalog dataset.
LLMs
MLPerf Inference uses:
Llama 3.1 405B with samples sourced from LongBench, LongDataCollection, RULER, and GovReport summary. Llama 2 70B uses OpenOrca. Llama 3.1 8B uses CNN/DailyMail. Mixtral 8x7B with samples sourced from OpenOrca, GSM8K, and MBXP datasets.
MLPerf Training uses:
The Llama 3.1 generative language model with 405 billion parameters and a sequence length of 8,192 for the LLM pretraining workload with the c4 (v3.0.1) dataset. For the LLM fine-tuning test, it uses the Llama 2 70B model with the GovReport dataset with sequence lengths of 8,192. Llama 3.1 8B also uses the C4 dataset with sequence lengths of 8,192..
Text-to-Video
MLPerf Inference uses Wan-2.2-T2V-A14B with the VBench dataset.
Text-to-Image
MLPerf Training uses the FLUX.1 text-to-image model trained on the CC12M dataset with the COCO 2014 dataset for eval.
Recommender
MLPerf Inference uses DLRMv3 with a Synthetic Streaming 100B dataset.
MLPerf Training and Inference use the Deep Learning Recommendation Model v2 (DLRMv2) that employs DCNv2 cross-layer and a multi-hot dataset synthesized from the Criteo dataset.
Graph Neural Network
MLPerf Inference uses the Illinois Graph Benchmark (IGB) heterogeneous dataset.
MLPerf Training uses R-GAT with the Illinois Graph Benchmark (IGB) heterogeneous dataset.
Speech-to-Text
MLPerf Inference uses Whisper-Large-V3 with the LibriSpeech dataset.
- Accelerated Apps Catalog
- Blackwell Resources Center
- Data Center GPUs
- Data Center GPU Line Card
- Data Center GPUs Resource Center
- Data Center Product Performance
- Deep Learning Institute
- Energy Efficiency Calculator
- GPU Cloud Computing
- MLPerf Benchmarks
- NGC Catalog
- NVIDIA-Certified Systems
- NVIDIA Data Center Corporate Blogs
- NVIDIA Data Center Technical Blogs
- Qualified System Catalog
- Where to Buy