MiniMax M2.5/M2.7 · GB200 NVL72 · Precision comparison

GB200 NVL72: FP4 vs FP8 precision comparison

How FP4 and FP8 precision affect MiniMax M2.5/M2.7 inference on GB200 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.

Throughput at 60 tok/s/user on MiniMax M2.5/M2.7 (GB200 NVL72): FP4 hits 10,256 tok/s/chip, FP8 hits 3,263. Costs per million tokens land at $0.050 and $0.16 respectively. FP4 delivers 3.1× the tok/s/chip on the same chip, which makes it 68% cheaper per token. The cost-throughput tradeoff from lower precision is only part of the picture — see the evaluation page for accuracy data.

Around the middle of the 31–147 tok/s/user interactivity band, at 89 tok/s/user on MiniMax M2.5/M2.7 (GB200 NVL72): FP4 runs 5,504 tok/s/chip at $0.094 per million tokens, FP8 runs 831 at $0.62. FP4 delivers 6.6× the tok/s/chip on the same chip, which makes it 85% cheaper per token. Precision changes affect both inference speed and model quality — consult the evaluation tab for accuracy benchmarks.

At 118 tok/s/user on MiniMax M2.5/M2.7 (GB200 NVL72), FP4 delivers 2,019 tok/s/chip at $0.26 per million tokens; FP8 delivers 331 tok/s/chip at $1.56. FP4 delivers 6.1× the tok/s/chip on the same chip, which makes it 84% cheaper per token. Lower-precision quantization trades model accuracy for throughput — check the evaluation page for quality impact. (Numbers reflect the default 1k/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)
FP410,256FP83,263
FP45,504FP8831
FP42,019FP8331
Cost ($/M tok)
FP4$0.050FP8$0.16
FP4$0.094FP8$0.62
FP4$0.26FP8$1.56
tok/s/MW
FP45,484,468FP81,744,903
FP42,943,092FP8444,236
FP41,079,723FP8177,148
Concurrency
FP4~1,024FP8~318
FP4~395FP8~32
FP4~124FP8~9
MiniMax M2.5/M2.7: FP4 versus FP8 throughput and cost at matched interactivity levels on GB200 NVL72
FP4 versus FP8 throughput and cost per million tokens on GB200 NVL72 for this comparison's canonical default workload
Open full-size image ↗