Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and MI355X (AMD CDNA 4) on MiniMax M3.1 Flash 428B. 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 MiniMax M3.1 Flash 428B at 137 tok/s/user: 61494 / 58896 tok/s/chip, $0.01 / $0.01 per million tokens. MI355X is 44% cheaper per token; B300 delivers 4% more tok/s/chip.
Around the middle of the 6β531 tok/s/user interactivity band, at 269 tok/s/user on MiniMax M3.1 Flash 428B: B300 runs 24166 tok/s/chip at $0.03/M tokens, MI355X runs 15324 at $0.03/M. B300 is 5% cheaper per token; B300 delivers 58% more tok/s/chip.
Setting 400 tok/s/user as the target on MiniMax M3.1 Flash 428B, B300 produces 10775 tok/s/chip ($0.06 per million tokens) and MI355X produces 7641 ($0.05). MI355X is 7% cheaper per token; B300 delivers 41% more tok/s/chip. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:61493.7MI355X:58896.1 | B300:24165.9MI355X:15323.6 | B300:10775.4MI355X:7640.9 |
| Cost ($/M tok) | B300:$0.010MI355X:$0.007 | B300:$0.026MI355X:$0.027 | B300:$0.058MI355X:$0.055 |
| tok/s/MW | B300:32365126MI355X:28179946 | B300:12718916MI355X:7331875 | B300:5671258MI355X:3655956 |
| Concurrency | B300:~42MI355X:~33 | B300:~13MI355X:~8 | B300:~5MI355X:~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.
