Head-to-head AI inference benchmark comparison of B300 (NVIDIA Blackwell) and GB300 NVL72 (NVIDIA Blackwell) on MiniMax M3 428B. 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 159 tok/s/user on MiniMax M3 428B: B300 hits 34927 tok/s/chip, GB300 NVL72 hits 32568. Per-million costs land at $0.02 and $0.02 respectively. B300 is 10% cheaper per token; B300 delivers 7% more tok/s/chip.
B300 / GB300 NVL72 on MiniMax M3 428B at 218 tok/s/user: 22298 / 21832 tok/s/chip, $0.03 / $0.03 per million tokens. B300 is 4% cheaper per token; B300 delivers 2% more tok/s/chip.
Toward the upper edge of the 100–336 tok/s/user interactivity band, at 277 tok/s/user on MiniMax M3 428B: B300 runs 9029 tok/s/chip at $0.07/M tokens, GB300 NVL72 runs 13103 at $0.05/M. GB300 NVL72 is 42% cheaper per token; GB300 NVL72 delivers 45% more tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | B300:34927.5GB300 NVL72:32567.6 | B300:22297.9GB300 NVL72:21832.5 | B300:9029.2GB300 NVL72:13103.4 |
| Cost ($/M tok) | B300:$0.018GB300 NVL72:$0.020 | B300:$0.028GB300 NVL72:$0.029 | B300:$0.070GB300 NVL72:$0.049 |
| tok/s/MW | B300:18382891GB300 NVL72:15362054 | B300:11735735GB300 NVL72:10298334 | B300:4752188GB300 NVL72:6180856 |
| Concurrency | B300:~20GB300 NVL72:~29 | B300:~15GB300 NVL72:~20 | B300:~5GB300 NVL72:~24 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
TCO $/chip/hr:
Source: SemiAnalysis Market July 2026 Pricing Surveys & AI Cloud TCO Model
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate cost per million tokens per decode chip or per prefill chip, rather than per total chip count. This makes direct token cost comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate input throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct input throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate output throughput per decode chip or per prefill chip, rather than per total chip count. This makes direct output throughput comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate power per decode chip or per prefill chip, rather than per total chip count. This makes direct power comparison with aggregated configs not an apples-to-apples comparison.
Note: Disaggregated inference configurations (e.g., MoRI SGLang, Dynamo TRTLLM) calculate Joules per decode chip or per prefill chip, rather than per total chip count. This makes direct Joules per token comparison with aggregated configs not an apples-to-apples comparison.
Matching measurements exist, but their chip series are hidden.
Shift+Scroll to zoom • Drag to pan • Double-click to reset • Click a point to pin tooltip
MiniMax M3 428B 8K / 1K Total Tokens per $1 TCO vs. Interactivity