DeepSeek V4 Pro 0813 1.6T · Chip comparison
DeepSeek V4 Pro 0813 1.6T — B200 vs GB300 NVL72
Head-to-head AI inference benchmark comparison of B200 (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 →
Throughput at 102 tok/s/user on DeepSeek V4 Pro 0813 1.6T: B200 hits 39,229 tok/s/chip, GB300 NVL72 hits 85,648. Costs per million tokens land at $0.012 and $0.007 respectively. GB300 NVL72 is 39% cheaper per token; GB300 NVL72 delivers 118% more tok/s/chip.
B200 / GB300 NVL72 on DeepSeek V4 Pro 0813 1.6T at 166 tok/s/user: 13,414 / 23,405 tok/s/chip, $0.036 / $0.027 per million tokens. GB300 NVL72 is 23% cheaper per token; GB300 NVL72 delivers 74% more tok/s/chip.
Toward the upper edge of the 38–294 tok/s/user interactivity band, at 231 tok/s/user on DeepSeek V4 Pro 0813 1.6T: B200 runs 4,040 tok/s/chip at $0.12 per million tokens, GB300 NVL72 runs 6,162 at $0.10. GB300 NVL72 is 12% cheaper per token; GB300 NVL72 delivers 53% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B20039,229GB300 NVL7285,648 | B20013,414GB300 NVL7223,405 | B2004,040GB300 NVL726,162 |
| Cost ($/M tok) | B200$0.012GB300 NVL72$0.007 | B200$0.036GB300 NVL72$0.027 | B200$0.12GB300 NVL72$0.10 |
| tok/s/MW | B20022,940,845GB300 NVL7240,399,857 | B2007,844,728GB300 NVL7211,039,976 | B2002,362,386GB300 NVL722,906,667 |
| Concurrency | B200~139GB300 NVL72~440 | B200~21GB300 NVL72~88 | B200~4GB300 NVL72~2 |