MiniMax M3 428B · B300 · Precision comparison

B300: FP4 vs FP8 precision comparison

How FP4 and FP8 precision affect MiniMax M3 428B inference on B300 (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.

At 122 tok/s/user on MiniMax M3 428B (B300), FP4 delivers 5,953 tok/s/chip at $0.11 per million tokens; FP8 delivers 3,849 tok/s/chip at $0.16. FP4 delivers 55% more tok/s/chip on the same chip, which makes it 35% cheaper per token. Lower-precision quantization trades model accuracy for throughput — check the evaluation page for quality impact.

FP4 posts 3,098 tok/s/chip for $0.20 per million tokens at 235 tok/s/user on MiniMax M3 428B (B300); FP8 posts 2,237 tok/s/chip for $0.28. FP4 delivers 39% more tok/s/chip on the same chip, which makes it 28% cheaper per token. Quantization-level accuracy differences are tracked on the evaluation tab.

Throughput at 349 tok/s/user on MiniMax M3 428B (B300): FP4 hits 2,067 tok/s/chip, FP8 hits 1,040. Costs per million tokens land at $0.30 and $0.60 respectively. FP4 delivers 99% more tok/s/chip on the same chip, which makes it 50% cheaper per token. The cost-throughput tradeoff from lower precision is only part of the picture — see the evaluation page for accuracy data. (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.)

Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.

Metric
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Interactivity (tok/s/user)
Throughput (tok/s/chip)
FP45,953FP83,849
FP43,098FP82,237
FP42,067FP81,040
Cost ($/M tok)
FP4$0.11FP8$0.16
FP4$0.20FP8$0.28
FP4$0.30FP8$0.60
tok/s/MW
FP43,133,311FP82,025,703
FP41,630,350FP81,177,121
FP41,087,633FP8547,556
Concurrency
FP4~12FP8~16
FP4~8FP8~4
FP4~3FP8~3
MiniMax M3 428B: FP4 versus FP8 throughput and cost at matched interactivity levels on B300
FP4 versus FP8 throughput and cost per million tokens on B300 for this comparison's canonical default workload
Open full-size image ↗