Qwen 3.5 397B-A17B · Chip comparison

Qwen 3.5 397B-A17B — MI300X vs MI325X

Head-to-head AI inference benchmark comparison of MI300X (AMD CDNA 3) 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 →

At 28 tok/s/user interactivity on Qwen 3.5 397B-A17B, MI300X delivers 12,524 tok/s/chip at $0.021 per million tokens; MI325X delivers 11,658 tok/s/chip at $0.026. MI300X is 20% cheaper per token; MI300X delivers 7% more tok/s/chip at this point.

MI300X posts 12,275 tok/s/chip for $0.021 per million tokens at 38 tok/s/user on Qwen 3.5 397B-A17B; MI325X posts 9,668 tok/s/chip for $0.032. MI300X is 32% cheaper per token; MI300X delivers 27% more tok/s/chip.

Throughput at 48 tok/s/user on Qwen 3.5 397B-A17B: MI300X hits 11,971 tok/s/chip, MI325X hits 7,468. Costs per million tokens land at $0.022 and $0.041 respectively. MI300X is 46% cheaper per token; MI300X delivers 60% 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 →

MI325X vs MI300X: 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)
MI300X12,524MI325X11,658
MI300X12,275MI325X9,668
MI300X11,971MI325X7,468
Cost ($/M tok)
MI300X$0.021MI325X$0.026
MI300X$0.021MI325X$0.032
MI300X$0.022MI325X$0.041
tok/s/MW
MI300X9,010,240MI325X6,898,455
MI300X8,831,044MI325X5,720,808
MI300X8,612,028MI325X4,419,157
Concurrency
MI300X~25MI325X~17
MI300X~24MI325X~20
MI300X~22MI325X~8