Qwen 3.5 397B-A17B · Chip comparison

Qwen 3.5 397B-A17B — H200 vs MI325X

Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and MI325X (AMD CDNA 3) 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 →

H200 / MI325X on Qwen 3.5 397B-A17B at 43 tok/s/user: 20,890 / 8,505 tok/s/chip, $0.016 / $0.036 per million tokens. H200 is 55% cheaper per token; H200 delivers 146% more tok/s/chip.

Around the middle of the 39–58 tok/s/user interactivity band, at 48 tok/s/user on Qwen 3.5 397B-A17B: H200 runs 20,723 tok/s/chip at $0.016 per million tokens, MI325X runs 7,468 at $0.041. H200 is 60% cheaper per token; H200 delivers 177% more tok/s/chip.

Setting 53 tok/s/user as the target on Qwen 3.5 397B-A17B, H200 produces 20,554 tok/s/chip ($0.016 per million tokens) and MI325X produces 6,596 ($0.046). H200 is 64% cheaper per token; H200 delivers 3.1× the 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 →

MI325X vs H200: 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)
H20020,890MI325X8,505
H20020,723MI325X7,468
H20020,554MI325X6,596
Cost ($/M tok)
H200$0.016MI325X$0.036
H200$0.016MI325X$0.041
H200$0.016MI325X$0.046
tok/s/MW
H20015,248,024MI325X5,032,512
H20015,126,600MI325X4,419,157
H20015,003,083MI325X3,903,073
Concurrency
H200~46MI325X~11
H200~44MI325X~8
H200~42MI325X~9