Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on Muse Spark Open 816B. 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
B300 / MI355X on Muse Spark Open 816B at 68 tok/s/user: 21347 / 8974 tok/s/chip, $0.03 / $0.05 per million tokens. B300 is 58% cheaper per token; B300 delivers 138% more tok/s/chip.
Around the middle of the 30β184 tok/s/user interactivity band, at 107 tok/s/user on Muse Spark Open 816B: B300 runs 21347 tok/s/chip at $0.03/M tokens, MI355X runs 5413 at $0.08/M. B300 is 162% cheaper per token; B300 delivers 294% more tok/s/chip.
Setting 145 tok/s/user as the target on Muse Spark Open 816B, B300 produces 17928 tok/s/chip ($0.04 per million tokens) and MI355X produces 2782 ($0.15). B300 is 328% cheaper per token; B300 delivers 544% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:21346.8MI355X:8974.0 | B300:21346.8MI355X:5412.7 | B300:17928.2MI355X:2781.8 |
| Cost ($/M tok) | B300:$0.029MI355X:$0.046 | B300:$0.029MI355X:$0.077 | B300:$0.035MI355X:$0.150 |
| tok/s/MW | B300:11235170MI355X:4293757 | B300:11235170MI355X:2589817 | B300:9435917MI355X:1330995 |
| Concurrency | B300:~16MI355X:~15 | B300:~16MI355X:~10 | B300:~14MI355X:~1 |
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.
