GB300 NVL72: FP4 vs FP8 Precision Comparison
How FP4 and FP8 precision affect MiniMax M3 428B inference on GB300 NVL72 (NVIDIA Blackwell). Throughput, latency, and cost across LLM workloads. Use the chart controls below to switch sequences and metrics — same interactions as the main inference chart.
FP8: 4610 tok/s/chip, $0.14 per million tokens at 125 tok/s/user on MiniMax M3 428B (GB300 NVL72). FP4 is unmeasured here.
At 229 tok/s/user on MiniMax M3 428B (GB300 NVL72), FP8 delivers 1782 tok/s/chip at $0.36 per million tokens; FP4 hasn't been benchmarked at this target.
FP8 hits 1030 tok/s/chip for $0.62 per million tokens at 333 tok/s/user on MiniMax M3 428B (GB300 NVL72). No FP4 data at this operating point. (Numbers reflect the default 8k/1k selection for this URL — table and chart below update if you change sequence or model in the controls. Each side uses the best available serving configuration for that precision, which may include speculative decoding such as MTP where recipes exist — the same convention as the other comparison pages.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | FP4:—FP8:4610.3 | FP4:—FP8:1782.4 | FP4:—FP8:1030.4 |
| Cost ($/M tok) | FP4:—FP8:$0.139 | FP4:—FP8:$0.360 | FP4:—FP8:$0.623 |
| tok/s/MW | FP4:—FP8:2174679 | FP4:—FP8:840769 | FP4:—FP8:486019 |
| Concurrency | FP4:—FP8:~60 | FP4:—FP8:~8 | FP4:—FP8:~4 |
Inference Performance
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.