Head-to-head AI inference benchmark comparison of MI300X (AMD CDNA 3) and MI355X (AMD CDNA 4) on DeepSeek V4.1 Flash 552B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics β 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 β
Throughput at 83 tok/s/user on DeepSeek V4.1 Flash 552B: MI300X hits 7758 tok/s/chip, MI355X hits 131853. Per-million costs land at $0.03 and $0.00 respectively. MI355X is 976% cheaper per token; MI355X delivers 1600% more tok/s/chip.
MI300X / MI355X on DeepSeek V4.1 Flash 552B at 123 tok/s/user: 4425 / 105169 tok/s/chip, $0.06 / $0.00 per million tokens. MI355X is 1405% cheaper per token; MI355X delivers 2277% more tok/s/chip.
Toward the upper edge of the 43β204 tok/s/user interactivity band, at 164 tok/s/user on DeepSeek V4.1 Flash 552B: MI300X runs 3722 tok/s/chip at $0.07/M tokens, MI355X runs 83368 at $0.00/M. MI355X is 1319% cheaper per token; MI355X delivers 2140% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI300X:7757.6MI355X:131853.0 | MI300X:4425.3MI355X:105169.3 | MI300X:3721.6MI355X:83367.8 |
| Cost ($/M tok) | MI300X:$0.034MI355X:$0.003 | MI300X:$0.060MI355X:$0.004 | MI300X:$0.071MI355X:$0.005 |
| tok/s/MW | MI300X:5580994MI355X:63087544 | MI300X:3183656MI355X:50320250 | MI300X:2677387MI355X:39888881 |
| Concurrency | MI300X:~6MI355X:~60 | MI300X:~3MI355X:~32 | MI300X:~1MI355X:~32 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
