Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) 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 210 tok/s/user on DeepSeek V4.1 Flash 552B: B300 hits 101219 tok/s/chip, MI355X hits 59646. Per-million costs land at $0.01 and $0.01 respectively. B300 is 13% cheaper per token; B300 delivers 70% more tok/s/chip.
B300 / MI355X on DeepSeek V4.1 Flash 552B at 354 tok/s/user: 40311 / 31817 tok/s/chip, $0.02 / $0.01 per million tokens. MI355X is 19% cheaper per token; B300 delivers 27% more tok/s/chip.
Toward the upper edge of the 67β641 tok/s/user interactivity band, at 497 tok/s/user on DeepSeek V4.1 Flash 552B: B300 runs 28649 tok/s/chip at $0.02/M tokens, MI355X runs 15332 at $0.03/M. B300 is 24% cheaper per token; B300 delivers 87% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:101219.1MI355X:59646.3 | B300:40311.3MI355X:31816.9 | B300:28649.3MI355X:15331.5 |
| Cost ($/M tok) | B300:$0.006MI355X:$0.007 | B300:$0.016MI355X:$0.013 | B300:$0.022MI355X:$0.027 |
| tok/s/MW | B300:53273202MI355X:28538912 | B300:21216459MI355X:15223413 | B300:15078553MI355X:7335665 |
| Concurrency | B300:~32MI355X:~30 | B300:~20MI355X:~8 | B300:~15MI355X:~2 |
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.
