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 — 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 28 tok/s/user on DeepSeek V4.1 Flash 552B: B300 hits 101808 tok/s/chip, MI355X hits 5021. Per-million costs land at $0.01 and $0.08 respectively. B300 is 1246% cheaper per token; B300 delivers 1928% more tok/s/chip.
B300 / MI355X on DeepSeek V4.1 Flash 552B at 31 tok/s/user: 100523 / 2980 tok/s/chip, $0.01 / $0.14 per million tokens. B300 is 2139% cheaper per token; B300 delivers 3273% more tok/s/chip.
Toward the upper edge of the 26–35 tok/s/user interactivity band, at 33 tok/s/user on DeepSeek V4.1 Flash 552B: B300 runs 99660 tok/s/chip at $0.01/M tokens, MI355X runs 2184 at $0.19/M. B300 is 2929% cheaper per token; B300 delivers 4464% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:101808.1MI355X:5020.7 | B300:100522.6MI355X:2980.1 | B300:99659.9MI355X:2183.7 |
| Cost ($/M tok) | B300:$0.006MI355X:$0.083 | B300:$0.006MI355X:$0.140 | B300:$0.006MI355X:$0.191 |
| tok/s/MW | B300:53583196MI355X:2402272 | B300:52906635MI355X:1425899 | B300:52452579MI355X:1044818 |
| Concurrency | B300:~124MI355X:~7 | B300:~116MI355X:~4 | B300:~111MI355X:~3 |
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.
