DeepSeek R1 · Chip comparison

DeepSeek R1 — B200 vs MI325X

Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI325X (AMD CDNA 3) 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.

Near the low end of the 14–100 tok/s/user interactivity band, at 35 tok/s/user on DeepSeek R1: B200 runs 5,654 tok/s/chip at $0.085 per million tokens, MI325X runs 966 at $0.32. B200 is 73% cheaper per token; B200 delivers 5.9× the tok/s/chip.

Setting 57 tok/s/user as the target on DeepSeek R1, B200 produces 2,917 tok/s/chip ($0.16 per million tokens) and MI325X produces 747 ($0.41). B200 is 60% cheaper per token; B200 delivers 3.9× the tok/s/chip.

At 78 tok/s/user interactivity on DeepSeek R1, B200 delivers 2,166 tok/s/chip at $0.22 per million tokens; MI325X delivers 607 tok/s/chip at $0.50. B200 is 56% cheaper per token; B200 delivers 3.6× the tok/s/chip at this point. (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.)

View performance per dollar →

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)
B2005,654MI325X966
B2002,917MI325X747
B2002,166MI325X607
Cost ($/M tok)
B200$0.085MI325X$0.32
B200$0.16MI325X$0.41
B200$0.22MI325X$0.50
tok/s/MW
B2003,306,371MI325X571,610
B2001,705,624MI325X442,003
B2001,266,497MI325X358,995
Concurrency
B200~712MI325X~24
B200~286MI325X~11
B200~104MI325X~7