DeepSeek V4.1 Flash 552B — MI300X vs MI355X
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 — 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 →
Throughput at 94 tok/s/user on DeepSeek V4.1 Flash 552B: MI300X hits 8308 tok/s/chip, MI355X hits 157490. Per-million costs land at $0.03 and $0.00 respectively. MI355X is 1101% cheaper per token; MI355X delivers 1796% more tok/s/chip.
MI300X / MI355X on DeepSeek V4.1 Flash 552B at 131 tok/s/user: 5659 / 118406 tok/s/chip, $0.05 / $0.00 per million tokens. MI355X is 1225% cheaper per token; MI355X delivers 1992% more tok/s/chip.
Toward the upper edge of the 57–205 tok/s/user interactivity band, at 168 tok/s/user on DeepSeek V4.1 Flash 552B: MI300X runs 3808 tok/s/chip at $0.07/M tokens, MI355X runs 109663 at $0.00/M. MI355X is 1724% cheaper per token; MI355X delivers 2780% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI300X:8308.3MI355X:157489.6 | MI300X:5659.3MI355X:118405.7 | MI300X:3808.4MI355X:109663.0 |
| Cost ($/M tok) | MI300X:$0.032MI355X:$0.003 | MI300X:$0.047MI355X:$0.004 | MI300X:$0.069MI355X:$0.004 |
| tok/s/MW | MI300X:5977189MI355X:75353896 | MI300X:4071419MI355X:56653430 | MI300X:2739848MI355X:52470354 |
| Concurrency | MI300X:~6MI355X:~51 | MI300X:~4MI355X:~33 | MI300X:~1MI355X:~32 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.