DeepSeek R1 · Chip comparison
DeepSeek R1 — B200 vs H100
Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and H100 (NVIDIA Hopper) on DeepSeek R1. 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.
B200 / H100 on DeepSeek R1 at 45 tok/s/user: 5,215 / 739 tok/s/chip, $0.092 / $0.44 per million tokens. B200 is 79% cheaper per token; B200 delivers 7.1× the tok/s/chip.
Around the middle of the 20–122 tok/s/user interactivity band, at 71 tok/s/user on DeepSeek R1: B200 runs 2,846 tok/s/chip at $0.17 per million tokens, H100 runs 281 at $1.16. B200 is 85% cheaper per token; B200 delivers 10× the tok/s/chip.
Setting 97 tok/s/user as the target on DeepSeek R1, B200 produces 1,429 tok/s/chip ($0.34 per million tokens) and H100 produces 154 ($2.11). B200 is 84% cheaper per token; B200 delivers 9.3× the tok/s/chip. (Numbers reflect the default 8k/1k · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B2005,215H100739 | B2002,846H100281 | B2001,429H100154 |
| Cost ($/M tok) | B200$0.092H100$0.44 | B200$0.17H100$1.16 | B200$0.34H100$2.11 |
| tok/s/MW | B2003,049,713H100539,778 | B2001,664,604H100204,852 | B200835,812H100112,644 |
| Concurrency | B200~365H100~132 | B200~219H100~41 | B200~118H100~17 |