B200: FP4 vs FP8 Precision Comparison

At 47 tok/s/user on MiniMax M2.5/M2.7 (B200), FP4 delivers 15557 tok/s/GPU at $0.03 per million tokens; FP8 delivers 9292 tok/s/GPU at $0.06. FP4 is 68% cheaper per token; FP4 delivers 67% more tok/s/GPU. Lower-precision quantization trades model accuracy for throughput — check the evaluation page for quality impact.

FP4 posts 11669 tok/s/GPU for $0.05 per million tokens at 75 tok/s/user on MiniMax M2.5/M2.7 (B200); FP8 posts 3367 tok/s/GPU for $0.16. FP4 is 249% cheaper per token; FP4 delivers 247% more tok/s/GPU. Quantization-level accuracy differences are tracked on the evaluation tab.

Throughput at 103 tok/s/user on MiniMax M2.5/M2.7 (B200): FP4 hits 6394 tok/s/GPU, FP8 hits 2070. Per-million costs land at $0.08 and $0.26 respectively. FP4 is 211% cheaper per token; FP4 delivers 209% more tok/s/GPU. 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.)