这是 NVIDIA 的数据中心深度学习产品性能中心，汇集可复现的 AI 性能基准数据，覆盖 NVIDIA 最新一代数据中心 GPU。
仅以原始运算速度作为唯一核心指标的时代已经过去。当下，行业更关注规模化场景下的吞吐量、能效与综合成本效益。随着 AI 从单次直接作答，逐步演进为支持多步推理，市场对推理能力及其底层经济性的需求持续上涨。这种转变会大幅提升算力需求，因为单次请求需要生成远更多 token。除吞吐量外，每瓦 token 数、每百万 token 成本、单用户每秒 token 数等指标同样至关重要。
对于受功耗约束的 AI 算力工厂，NVIDIA 持续迭代的软件优化能够长期提升 token 产出收益，凸显了我们技术创新的价值。
帕累托曲线 (Pareto curves) 直观展现，NVIDIA Blackwell 架构可在各类生产核心目标间实现最优均衡，包含：
成本
能效
吞吐量
响应速度
若仅针对单一场景优化系统，会限制部署灵活性，导致在曲线其他工况下效率不足。NVIDIA 的全栈设计思路，可在多种真实生产场景中兼顾能效与收益。Blackwell 的领先性源自极致的软硬件协同设计，这套全栈架构专为高性能、高能效与高可扩展性打造。
| Network | Throughput | GPU | Server | GPU Version | QSL Size | Target Accuracy | Dataset |
|---|---|---|---|---|---|---|---|
| DeepSeek R1 | 2,494,310 tokens/sec | 288x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 4388 | 99% of FP16 (exact match 81.9132%) | mlperf_deepseek_r1 |
| 486,141 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 4388 | 99% of FP16 (exact match 81.9132%) | mlperf_deepseek_r1 | |
| 70,326 tokens/sec | 8x B300 | NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 4388 | 99% of FP16 (exact match 81.9132%) | mlperf_deepseek_r1 | |
| 58,582 tokens/sec | 8x B200 | Nebius B200 n1 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 4388 | 99% of FP16 (exact match 81.9132%) | mlperf_deepseek_r1 | |
| gpt-oss 120B | 1,046,150 tokens/sec | 72x GB300 | Nebius GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 6396 | 99% of 83.13% | AIME25, GPQA Diamond, LiveCodeBench v6 |
| 879,542 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 6396 | 99% of 83.13% | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| 111,496 tokens/sec | 8x B300 | Cisco UCS C880A M8 (8x NVIDIA B300-SXM-270GB, TensorRT) | NVIDIA B300 | 6396 | 99% of 83.13% | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| 93,071 tokens/sec | 8x B200 | LLM-D v0.5.0,Openshift 4.20.12,NVIDIA 8xB200-SXM-180GB | NVIDIA B200 | 6396 | 99% of 83.13% | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| Qwen3-VL 235B | 61 tokens/sec | 4x GB300 | NVIDIA GB300 NVL72 (4x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | Shopify Product Catalogue |
| 44 tokens/sec | 4x GB200 | NVIDIA GB200 NVL72 (4x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | Shopify Product Catalogue | |
| 78 tokens/sec | 8x B300 | Nebius B300 n1 (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | Shopify Product Catalogue | |
| 79 tokens/sec | 8x B200 | Dell B200,8xB200-SXM-180GB,RHEL 10.1,vLLM CentML:mlperf-inf-mm-q3vl-v6.0 | NVIDIA B200 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | Shopify Product Catalogue | |
| Llama3.1 405B | 19,512 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | Subset of LongBench, LongDataCollections, Ruler, GovReport |
| 15,462 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| 1,971 tokens/sec | 8x B300 | Cisco UCS C880A M8 (8x NVIDIA B300-SXM-270GB, TensorRT) | NVIDIA B300 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| 1,350 tokens/sec | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| Llama2 70B | 1,126,850 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | OpenOrca (max_seq_len=1024) |
| 888,054 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | OpenOrca (max_seq_len=1024) | |
| 112,954 tokens/sec | 8x B300 | NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | OpenOrca (max_seq_len=1024) | |
| 104,572 tokens/sec | 8x B200 | HPE ProLiant Compute XD685 (8x NVIDIA B200 180GB, TensorRT) | NVIDIA B200 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | OpenOrca (max_seq_len=1024) | |
| Llama3.1 8B | 166,745 tokens/sec | 8x B300 | XA NB3I-E12 | NVIDIA B300 | 13368 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881). Additionally, for both cases the total generation length of the texts should be more than 90% of the reference (gen_len=8167644) | CNN Dailymail (v3.0.0, max_seq_len=2048) |
| 160,403 tokens/sec | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 13368 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881). Additionally, for both cases the total generation length of the texts should be more than 90% of the reference (gen_len=8167644) | CNN Dailymail (v3.0.0, max_seq_len=2048) | |
| Wan2.2 | 0.037 samples/sec | 4x GB300 | NVIDIA GB300 NVL72 (4x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 248 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881) | VBench prompts |
| 0.027 samples/sec | 4x GB200 | NVIDIA GB200 NVL72 (4x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 248 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881) | VBench prompts | |
| 0.059 samples/sec | 8x B300 | NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 248 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881) | VBench prompts | |
| 0.046 samples/sec | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 248 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881) | VBench prompts | |
| DLRMv3 | 104,637 samples/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 34996 | 99% of FP32 and 99.9% of FP32 (AUC=80.31%) | Synthetic Streaming 100B Dataset |
| 10,737 samples/sec | 8x B200 | Camarero PDI200A2HG-810 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 34996 | 99% of FP32 and 99.9% of FP32 (WER=2.0671%) | Synthetic Streaming 100B Dataset | |
| Whisper | 50,562 samples/sec | 8x B300 | NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 1633 | 99% of FP32 and 99.9% of FP32 (WER=2.0671%) | LibriSpeech |
| 49,327 samples/sec | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 1633 | 99% of FP32 and 99.9% of FP32 (WER=2.0671%) | LibriSpeech |
| Network | Throughput | GPU | Server | GPU Version | QSL Size | Target Accuracy | MLPerf Server Latency
Constraints (ms) |
Dataset |
|---|---|---|---|---|---|---|---|---|
| DeepSeek R1 | 1,555,110 tokens/sec | 288x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 4388 | 99% of FP16 (exact match 81.9132%) | TTFT/TPOT: 2000 ms/80 ms | mlperf_deepseek_r1 |
| 336,106 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 4388 | 99% of FP16 (exact match 81.9132%) | TTFT/TPOT: 2000 ms/80 ms | mlperf_deepseek_r1 | |
| 60,413 tokens/sec | 8x B300 | Nebius B300 n1 (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 4388 | 99% of FP16 (exact match 81.9132%) | TTFT/TPOT: 2000 ms/80 ms | mlperf_deepseek_r1 | |
| 51,693 tokens/sec | 8x B200 | Nebius B200 n1 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 4388 | 99% of FP16 (exact match 81.9132%) | TTFT/TPOT: 2000 ms/80 ms | mlperf_deepseek_r1 | |
| gpt-oss 120B | 1,096,770 tokens/sec | 72x GB300 | Nebius GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 6396 | 99% of 83.13% | TTFT/TPOT: 3000 ms/80 ms | AIME25, GPQA Diamond, LiveCodeBench v6 |
| 899,218 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 6396 | 99% of 83.13% | TTFT/TPOT: 3000 ms/80 ms | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| 110,655 queries/sec | 8x B300 | Cisco UCS C880A M8 (8x NVIDIA B300-SXM-270GB, TensorRT) | NVIDIA B300 | 6396 | 99% of 83.13% | TTFT/TPOT: 3000 ms/80 ms | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| 87,444 tokens/sec | 8x B200 | Nebius B200 n1 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 6396 | 99% of 83.13% | TTFT/TPOT: 3000 ms/80 ms | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| Qwen3-VL 235B | 43 tokens/sec | 4x GB300 | Nebius GB300 NVL72 (4x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | 12 s | Shopify Product Catalogue |
| 38 tokens/sec | 4x GB200 | NVIDIA GB200 NVL72 (4x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | 12 s | Shopify Product Catalogue | |
| 45 queries/sec | 8x B300 | Nebius B300 n1 (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | 12 s | Shopify Product Catalogue | |
| 68 tokens/sec | 8x B200 | Dell B200,8xB200-SXM-180GB,RHEL 10.1,vLLM CentML:mlperf-inf-mm-q3vl-v6.0 | NVIDIA B200 | 48289 | 99% of BF16 (Category Hierarchical F1 Score >= 0.7824) | 12 s | Shopify Product Catalogue | |
| Llama3.1 405B | 18,628 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | TTFT/TPOT: 6000 ms/175 ms | Subset of LongBench, LongDataCollections, Ruler, GovReport |
| 14,134 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | TTFT/TPOT: 6000 ms/175 ms | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| 1,484 tokens/sec | 8x B300 | QuantaGrid D75H-10U (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | TTFT/TPOT: 6000 ms/175 ms | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| 984 tokens/sec | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | TTFT/TPOT: 6000 ms/175 ms | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| Llama2 70B | 868,278 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 2000 ms/200 ms | OpenOrca (max_seq_len=1024) |
| 810,104 tokens/sec | 72x B200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 2000 ms/200 ms | OpenOrca (max_seq_len=1024) | |
| 108,392 tokens/sec | 8x B300 | PowerEdge XE9780L (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 2000 ms/200 ms | OpenOrca (max_seq_len=1024) | |
| 103,627 tokens/sec | 8x B200 | HPE ProLiant Compute XD685 (8x NVIDIA B200 180GB, TensorRT) | NVIDIA B200 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 2000 ms/200 ms | OpenOrca (max_seq_len=1024) | |
| Llama3.1 8B | 148,067 tokens/sec | 8x B300 | XA NB3I-E12 | NVIDIA B300 | 13368 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881). Additionally, for both cases the total generation length of the texts should be more than 90% of the reference (gen_len=8167644) | TTFT/TPOT: 2000 ms/100 ms | CNN Dailymail (v3.0.0, max_seq_len=2048) |
| 131,270 queries/sec | 8x B200 | HPE ProLiant Compute XD685 (8x NVIDIA B200 180GB, TensorRT) | NVIDIA B200 | 13368 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881). Additionally, for both cases the total generation length of the texts should be more than 90% of the reference (gen_len=8167644) | TTFT/TPOT: 2000 ms/100 ms | CNN Dailymail (v3.0.0, max_seq_len=2048) | |
| Wan2.2** | 31 seconds | 4x GB300 | NVIDIA GB300 NVL72 (4x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 248 | 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162) | N/A | VBench prompts |
| 40 seconds | 4x GB200 | NVIDIA GB200 NVL72 (4x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 248 | 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162) | N/A | VBench prompts | |
| 21 seconds | 8x B300 | G894-SD3-AAX7 | NVIDIA B300 | 248 | FID range: [23.01085758, 23.95007626] and CLIP range: [31.68631873, 31.81331801] | N/A | VBench prompts | |
| 25 seconds | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 248 | 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162) | N/A | VBench prompts | |
| DLRMv3 | 99,997 queries/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 34996 | 99% of FP32 (AUC=80.31%) | 80 ms | Synthetic Streaming 100B Dataset |
| 10,007 queries/sec | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 34996 | FID range: [23.01085758, 23.95007626] and CLIP range: [31.68631873, 31.81331801] | 80 ms | Synthetic Streaming 100B Dataset |
| Network | Throughput | GPU | Server | GPU Version | QSL Size | Target Accuracy | MLPerf Server Latency
Constraints (ms) |
Dataset |
|---|---|---|---|---|---|---|---|---|
| DeepSeek R1 | 250,634 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 4388 | 99% of FP16 (exact match 81.9132%) | TTFT/TPOT: 1500 ms/15 ms | mlperf_deepseek_r1 |
| 240,318 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 4388 | 99% of FP16 (exact match 81.9132%) | TTFT/TPOT: 1500 ms/15 ms | mlperf_deepseek_r1 | |
| 4,935 tokens/sec | 8x B300 | G894-SD3-AAX7 | NVIDIA B300 | 4388 | 99% of FP16 (exact match 81.9132%) | TTFT/TPOT: 1500 ms/15 ms | mlperf_deepseek_r1 | |
| gpt-oss 120B | 677,199 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 6396 | 99% of 83.13% | TTFT/TPOT: 2000 ms/20 ms | AIME25, GPQA Diamond, LiveCodeBench v6 |
| 624,929 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 6396 | 99% of 83.13% | TTFT/TPOT: 2000 ms/20 ms | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| 26,006 tokens/sec | 8x B300 | XA NB3I-E12 | NVIDIA B300 | 6396 | 99% of 83.13% | TTFT/TPOT: 2000 ms/20 ms | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| 13,155 tokens/sec | 8x B200 | Nebius B200 n1 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 6396 | 99% of 83.13% | TTFT/TPOT: 2000 ms/20 ms | AIME25, GPQA Diamond, LiveCodeBench v6 | |
| Llama3.1 405B | 18,365 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | TTFT/TPOT: 4500 ms/80 ms | Subset of LongBench, LongDataCollections, Ruler, GovReport |
| 14,010 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | TTFT/TPOT: 4500 ms/80 ms | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| 765 tokens/sec | 8x B300 | G894-SD3-AAX7 | NVIDIA B300 | 8313 | 99% of FP16 ((GovReport + LongDataCollections + 65 Sample from LongBench)rougeL=21.6666, (Remaining samples of the dataset)exact_match=90.1335). Additionally, for both cases tokens per sample should be between than 90% and 110% of the reference (tokens_per_sample=684.68) | TTFT/TPOT: 4500 ms/80 ms | Subset of LongBench, LongDataCollections, Ruler, GovReport | |
| Llama2 70B | 814,128 tokens/sec | 72x GB300 | NVIDIA GB300 NVL72 (72x GB300-288GB_aarch64, TensorRT) | NVIDIA GB300 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 450 ms/40 ms | OpenOrca (max_seq_len=1024) |
| 754,855 tokens/sec | 72x GB200 | NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT) | NVIDIA GB200 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 450 ms/40 ms | OpenOrca (max_seq_len=1024) | |
| 70,724 tokens/sec | 8x B300 | PowerEdge XE9780L (8x B300-SXM-270GB, TensorRT) | NVIDIA B300 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 450 ms/40 ms | OpenOrca (max_seq_len=1024) | |
| 61,300 tokens/sec | 8x B200 | HPE ProLiant Compute XD685 (8x NVIDIA B200 180GB, TensorRT) | NVIDIA B200 | 24576 | 99% of FP32 and 99.9% of FP32 (rouge1=44.4312, rouge2=22.0352, rougeL=28.6162). Additionally, for both cases the generation length of the tokens per sample should be more than 90% of the reference (tokens_per_sample=294.45) | TTFT/TPOT: 450 ms/40 ms | OpenOrca (max_seq_len=1024) | |
| Llama3.1 8B | 128,633 tokens/sec | 8x B300 | G894-SD3-AAX7 | NVIDIA B300 | 13368 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881). Additionally, for both cases the total generation length of the texts should be more than 90% of the reference (gen_len=8167644) | TTFT/TPOT: 500 ms/30 ms | CNN Dailymail (v3.0.0, max_seq_len=2048) |
| 128,750 tokens/sec | 8x B200 | NVIDIA DGX B200 (8x B200-SXM-180GB, TensorRT) | NVIDIA B200 | 13368 | 99% of FP32 and 99.9% of FP32 (rouge1=42.9865, rouge2=20.1235, rougeL=29.9881). Additionally, for both cases the total generation length of the texts should be more than 90% of the reference (gen_len=8167644) | TTFT/TPOT: 500 ms/30 ms | CNN Dailymail (v3.0.0, max_seq_len=2048) |
**The primary metric on Wan2.2 in Server Scenario is measured in seconds (lower the better).
MLPerf™ v6.0 Inference Closed Division. NVIDIA platform results from the following entries: 6.0-0006, 6.0-0010, 6.0-0024, 6.0-0039, 6.0-0040, 6.0-0048, 6.0-0062, 6.0-0072, 6.0-0073, 6.0-0074, 6.0-0075, 6.0-0076, 6.0-0077, 6.0-0078, 6.0-0080, 6.0-0081, 6.0-0083, 6.0-0084, 6.0-0085, 6.0-0089, 6.0-0091, 6.0-0094, 6.0-0098. MLPerf name and logo are trademarks. See
https://mlcommons.org/ for more information.
For MLPerf™ various scenario data, click
here
For MLPerf™ latency constraints, click
here
| Network | Batch Size | Throughput | Efficiency | Latency (ms) | GPU | Server | Container | Precision | Dataset | Framework | GPU Version |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Stable Video Diffusion | 1 | 7.32 videos/min | - | 8202.75 | 1x B200 | DGX B200 | 26.02-py3 | Mixed | Synthetic | TensorRT 10.15.1 | NVIDIA B200 |
| Stable Diffusion XL | 1 | 2.89 images/sec | - | 507.41 | 1x B200 | DGX B200 | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | NVIDIA B200 |
| BEVFusion Head | 1 | 2,543 images/sec | 6.03 images/sec/watt | 0.39 | 1x B200 | DGX B200 | 26.06-py3 | INT8 | Synthetic | TensorRT 11.0.0.114 | NVIDIA B200 |
| Flux Image Generator | 1 | 0.47 images/sec | - | 2130.4 | 1x B200 | DGX B200 | 26.02-py3 | FP4 | Synthetic | TensorRT 10.15.1 | NVIDIA B200 |
| HF Swin Base | 128 | 5,220 samples/sec | 5.69 samples/sec/watt | 24.52 | 1x B200 | DGX B200 | 26.04-py3 | FP8 | Synthetic | TensorRT 10.16.1.11 | NVIDIA B200 |
| HF Swin Large | 64 | 3,305 samples/sec | 3.38 samples/sec/watt | 19.36 | 1x B200 | DGX B200 | 26.04-py3 | FP8 | Synthetic | TensorRT 10.16.1.11 | NVIDIA B200 |
| HF ViT Base | 512 | 9,879 samples/sec | 10.38 samples/sec/watt | 51.83 | 1x B200 | DGX B200 | 26.04-py3 | FP8 | Synthetic | TensorRT 10.16.1.11 | NVIDIA B200 |
| HF ViT Large | 512 | 3,399 samples/sec | 3.55 samples/sec/watt | 301.21 | 1x B200 | DGX B200 | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | NVIDIA B200 |
| Yolo v10 M | 1 | 870 images/sec | 1.16 images/sec/watt | 1.15 | 1x B200 | DGX B200 | 26.06-py3 | INT8 | Synthetic | TensorRT 11.0.0.114 | NVIDIA B200 |
| Yolo v11 M | 1 | 1,069 images/sec | 1.38 images/sec/watt | 0.94 | 1x B200 | DGX B200 | 26.06-py3 | INT8 | Synthetic | TensorRT 11.0.0.114 | NVIDIA B200 |
HF Swin Base, HF Swin Large, HF ViT Base, HF ViT Large Sequence Length = 384
| Network | Batch Size | Throughput | Efficiency | Latency (ms) | GPU | Server | Container | Precision | Dataset | Framework | GPU Version |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Stable Diffusion XL | 1 | 1.05 images/sec | - | 954 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.01-py3 | FP8 | Synthetic | TensorRT 10.14.1 | RTX PRO 6000 BSE |
| Flux Image Generator | 1 | 0.2 images/sec | - | 5072 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.01-py3 | FP4 | Synthetic | TensorRT 10.14.1 | RTX PRO 6000 BSE |
| BEVFusion Head | 1 | 1738.51 images/sec | 5 images/sec/watt | 0.58 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | RTX PRO 6000 BSE |
| HF Swin Base | 32 | 2,719 samples/sec | 5 samples/sec/watt | 11.77 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | RTX PRO 6000 BSE |
| HF Swin Large | 32 | 1,517 samples/sec | 3 samples/sec/watt | 21.1 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | RTX PRO 6000 BSE |
| HF ViT Base | 512 | 3,419 samples/sec | 5.71 samples/sec/watt | 149.732 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | RTX PRO 6000 BSE |
| HF ViT Large | 512 | 1,161 samples/sec | 1.93 samples/sec/watt | 440.78 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | RTX PRO 6000 BSE |
| Yolo v11 M | 1 | 465 images/sec | 1 images/sec/watt | 2.15 | 1x RTX PRO 6000 | Supermicro SYS-521GE-TNRT | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | RTX PRO 6000 BSE |
HF Swin Base, HF Swin Large, HF ViT Base, HF ViT Large Sequence Length = 384
| Network | Batch Size | Throughput | Efficiency | Latency (ms) | GPU | Server | Container | Precision | Dataset | Framework | GPU Version |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Stable Diffusion XL | 1 | 0.4 images/sec | - | 2514 | 1x RTX PRO 4500 | Supermicro SYS-521GE-TNRT | 26.01-py3 | FP8 | Synthetic | TensorRT 10.14.1 | RTX PRO 4500 BSE |
| Flux Image Generator | 1 | 0.07 images/sec | - | 13816 | 1x RTX PRO 4500 | Supermicro SYS-521GE-TNRT | 26.01-py3 | FP4 | Synthetic | TensorRT 10.14.1 | RTX PRO 4500 BSE |
| HF Bert Large QAT | 64 | 2,720 samples/sec | - | 24 | 1x RTX PRO 4500 | Supermicro SYS-521GE-TNRT | 26.01-py3 | INT8 | Synthetic | TensorRT 10.14.1 | RTX PRO 4500 BSE |
| HF Bert Large | 64 | 1,507 samples/sec | - | 42 | 1x RTX PRO 4500 | Supermicro SYS-521GE-TNRT | 26.01-py3 | Mixed | Synthetic | TensorRT 10.14.1 | RTX PRO 4500 BSE |
| HF ViT Base | 16 | 1,403 samples/sec | - | 11 | 1x RTX PRO 4500 | Supermicro SYS-521GE-TNRT | 26.01-py3 | FP8 | Synthetic | TensorRT 10.14.1 | RTX PRO 4500 BSE |
| HF ViT Large | 4 | 449 samples/sec | - | 9 | 1x RTX PRO 4500 | Supermicro SYS-521GE-TNRT | 26.01-py3 | FP8 | Synthetic | TensorRT 10.14.1 | RTX PRO 4500 BSE |
HF Swin Base, HF Swin Large, HF ViT Base, HF ViT Large Sequence Length = 384
| Network | Batch Size | Throughput | Efficiency | Latency (ms) | GPU | Server | Container | Precision | Dataset | Framework | GPU Version |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Stable Diffusion XL | 1 | 1.54 images/sec | - | 780.31 | 1x H100 | DGX H100 | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | H100 SXM5-80GB |
| BEVFusion Head | 1 | 2,016 images/sec | 6.13 images/sec/watt | 0.5 | 1x H100 | DGX H100 | 26.06-py3 | INT8 | Synthetic | TensorRT 11.0.0.114 | H100 SXM5-80GB |
| HF Swin Base | 128 | 2,967 samples/sec | 4.29 samples/sec/watt | 43.14 | 1x H100 | DGX H100 | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | H100 SXM5-80GB |
| HF Swin Large | 128 | 1,831 samples/sec | 2.65 samples/sec/watt | 69.91 | 1x H100 | DGX H100 | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | H100 SXM5-80GB |
| HF ViT Base | 2048 | 4,939 samples/sec | 7.11 samples/sec/watt | 414.68 | 1x H100 | DGX H100 | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | H100 SXM5-80GB |
| HF ViT Large | 512 | 1,737 samples/sec | 7.54 samples/sec/watt | 294.79 | 1x H100 | DGX H100 | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | H100 SXM5-80GB |
| Yolo v10 M | 1 | 405 images/sec | 0.68 images/sec/watt | 2.47 | 1x H100 | DGX H100 | 26.04-py3 | FP8 | Synthetic | TensorRT 10.16.1.11 | H100 SXM5-80GB |
| Yolo v11 M | 1 | 480 images/sec | 0.76 images/sec/watt | 2.08 | 1x H100 | DGX H100 | 26.04-py3 | FP8 | Synthetic | TensorRT 10.16.1.11 | H100 SXM5-80GB |
HF Swin Base, HF Swin Large, HF ViT Base, HF ViT Large Sequence Length = 384
| Network | Batch Size | Throughput | Efficiency | Latency (ms) | GPU | Server | Container | Precision | Dataset | Framework | GPU Version |
|---|---|---|---|---|---|---|---|---|---|---|---|
| BEVFusion Head | 1 | 1958 images/sec | 7 images/sec/watt | 0.51 | 1x L40S | Supermicro SYS-521GE-TNRT | 26.02-py3 | INT8 | Synthetic | TensorRT 10.15.1 | NVIDIA L40S |
| HF Swin Base | 32 | 1,396 samples/sec | 4 samples/sec/watt | 22.92 | 1x L40S | Supermicro SYS-521GE-TNRT | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | NVIDIA L40S |
| HF Swin Large | 32 | 716 samples/sec | 2 samples/sec/watt | 44.72 | 1x L40S | Supermicro SYS-521GE-TNRT | 26.02-py3 | FP8 | Synthetic | TensorRT 10.15.1 | NVIDIA L40S |
| HF ViT Base | 1024 | 1,629 samples/sec | 4.76 samples/sec/watt | 628.73 | 1x L40S | Supermicro SYS-521GE-TNRT | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | NVIDIA L40S |
| HF ViT Large | 2048 | 578 samples/sec | 1.7 samples/sec/watt | 3,546.06 | 1x L40S | Supermicro SYS-521GE-TNRT | 26.06-py3 | FP8 | Synthetic | TensorRT 11.0.0.114 | NVIDIA L40S |
| Yolo v10 M | 1 | 275 images/sec | 0.79 images/sec/watt | 3.64 | 1x L40S | Supermicro SYS-521GE-TNRT | 26.02-py3 | INT8 | Synthetic | TensorRT 10.15.1 | NVIDIA L40S |
| Yolo v11 M | 1 | 310 images/sec | 0.9 images/sec/watt | 3.23 | 1x L40S | Supermicro SYS-521GE-TNRT | 26.02-py3 | INT8 | Synthetic | TensorRT 10.15.1 | NVIDIA L40S |
HF Swin Base, HF Swin Large, HF ViT Base, HF ViT Large Sequence Length = 384
只看计算单价或 FLOPs per dollar 会对推理 TCO 形成不完整的认知。对于 AI 推理 TCO，最重要的指标是每个 token 的成本，也就是实际交付的性价比。GB300 NVL72 在使用 Dynamo 和 TensorRT-LLM、单用户交互吞吐 116 TPS 的场景下，实现了每百万个 token 0.123 美元的推理成本——根据截至 2026 年 4 月的 SemiAnalysis InferenceX 基准测试，这是各大平台中每 token 成本最低的水平。
Metric |
NVIDIA Hopper (HGX H200) |
NVIDIA Blackwell (GB300 NVL72) |
NVIDIA Blackwell 相对 Hopper 的倍数 |
|---|---|---|---|
单 GPU 每小时成本 (美元) |
$1.41 |
$2.65 |
2 倍 |
每美元 FLOP (PFLOPS) |
2.8 |
5.6 |
2 倍 |
单 GPU 每秒 token 数 |
90 |
6,000 |
65 倍 |
每兆瓦每秒 token 数 |
54K |
2.8M |
50 倍 |
每百万 token 成本 (美元) |
$4.20 |
$0.12 |
降低 35 倍 |
GB300 NVL72 在使用 Dynamo 和 TensorRT-LLM、单用户交互吞吐 116 TPS 的场景下，实现了每百万个 token 0.123 美元的推理成本——根据截至 2026 年 4 月的 SemiAnalysis InferenceX 基准测试，这是各大平台中每 token 成本最低的水平。
NVIDIA 每百万 token 的推理成本在各代 GPU 中有了显著改善：根据 2026 年第一季度的 SemiAnalysis InferenceX 基准测试，在低延迟 agentic 工作负载上，得益于软硬件协同设计，NVIDIA Blackwell Ultra（GB300 NVL72）相较 NVIDIA Hopper 实现了每 MW 吞吐最高提升至 50 倍、每 token 成本最多降低至 35 倍。软件优化也带来持续改进——GB200 的 token 产出在三个月内提升了 4 倍，对应地每 token 成本也按比例下降。
NVIDIA 的 TensorRT-LLM 和 Dynamo 软件栈在无需更换硬件的前提下，持续带来推理成本优化。根据截至 2026 年 4 月的 SemiAnalysis InferenceX 基准测试，NVIDIA Blackwell B200 在 GPT-OSS-120B 模型上的每百万 token 成本，从发布时的 0.11 美元在两个月内降至 0.02 美元，单靠软件就实现了约 5 倍的改进。每个版本的 TensorRT-LLM 通常会通过算子/内核融合、量化改进以及调度优化等方式提升吞吐，从而进一步摊薄单位 token 的推理成本。