Qwen 3.5 397B-A17B · Chip comparison

Qwen 3.5 397B-A17B — B200 vs MI355X

Head-to-head AI inference benchmark comparison of B200 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on Qwen 3.5 397B-A17B. 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 →

At 113 tok/s/user interactivity on Qwen 3.5 397B-A17B, B200 delivers 34,902 tok/s/chip at $0.014 per million tokens; MI355X delivers 36,186 tok/s/chip at $0.012. MI355X is 16% cheaper per token; MI355X delivers 4% more tok/s/chip at this point.

B200 posts 20,578 tok/s/chip for $0.023 per million tokens at 196 tok/s/user on Qwen 3.5 397B-A17B; MI355X posts 18,291 tok/s/chip for $0.023. MI355X is 2% cheaper per token; B200 delivers 13% more tok/s/chip.

Throughput at 278 tok/s/user on Qwen 3.5 397B-A17B: B200 hits 9,469 tok/s/chip, MI355X hits 8,683. Costs per million tokens land at $0.051 and $0.048 respectively. MI355X is 5% cheaper per token; B200 delivers 9% more tok/s/chip. (Numbers reflect the default agentic-traces · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)

View performance per dollar →

MI355X vs B200: 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)
B20034,902MI355X36,186
B20020,578MI355X18,291
B2009,469MI355X8,683
Cost ($/M tok)
B200$0.014MI355X$0.012
B200$0.023MI355X$0.023
B200$0.051MI355X$0.048
tok/s/MW
B20020,410,757MI355X17,313,647
B20012,034,187MI355X8,751,732
B2005,537,582MI355X4,154,317
Concurrency
B200~63MI355X~29
B200~31MI355X~12
B200~12MI355X~6