DeepSeek V4 Pro 0813 1.6T · Chip comparison

DeepSeek V4 Pro 0813 1.6T — GB200 NVL72 vs GB300 NVL72

Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on DeepSeek V4 Pro 0813 1.6T. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics — same interactions as the main inference chart.

AgentX replays real coding-agent sessions rather than fixed-length prompts, so context grows turn over turn and most of each request is served from cache instead of being recomputed. That turns the comparison into a systems question: KV transfer between nodes, prefix-aware routing, and cache capacity all move the curve alongside raw chip throughput. Fixed-sequence workloads stay the clean baseline for kernel and silicon performance, so the two scenarios answer different questions about the same hardware. Learn more about AgentX →

Near the low end of the 56–287 tok/s/user interactivity band, at 113 tok/s/user on DeepSeek V4 Pro 0813 1.6T: GB200 NVL72 runs 30,877 tok/s/chip at $0.017 per million tokens, GB300 NVL72 runs 75,229 at $0.009. GB300 NVL72 is 49% cheaper per token; GB300 NVL72 delivers 144% more tok/s/chip.

Setting 171 tok/s/user as the target on DeepSeek V4 Pro 0813 1.6T, GB200 NVL72 produces 12,467 tok/s/chip ($0.041 per million tokens) and GB300 NVL72 produces 20,188 ($0.032). GB300 NVL72 is 23% cheaper per token; GB300 NVL72 delivers 62% more tok/s/chip.

At 229 tok/s/user interactivity on DeepSeek V4 Pro 0813 1.6T, GB200 NVL72 delivers 4,622 tok/s/chip at $0.11 per million tokens; GB300 NVL72 delivers 6,263 tok/s/chip at $0.10. GB300 NVL72 is 8% cheaper per token; GB300 NVL72 delivers 36% more tok/s/chip at this point. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)

View performance per dollar →

GB300 NVL72 vs GB200 NVL72: full specs & benchmarks across models →

Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.

Metric
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Throughput (tok/s/chip)
GB200 NVL7230,877GB300 NVL7275,229
GB200 NVL7212,467GB300 NVL7220,188
GB200 NVL724,622GB300 NVL726,263
Cost ($/M tok)
GB200 NVL72$0.017GB300 NVL72$0.009
GB200 NVL72$0.041GB300 NVL72$0.032
GB200 NVL72$0.11GB300 NVL72$0.10
tok/s/MW
GB200 NVL7216,511,700GB300 NVL7235,485,228
GB200 NVL726,666,914GB300 NVL729,522,504
GB200 NVL722,471,559GB300 NVL722,954,317
Concurrency
GB200 NVL72~198GB300 NVL72~355
GB200 NVL72~41GB300 NVL72~71
GB200 NVL72~6GB300 NVL72~2