B200: FP4 vs FP8 Precision Comparison

At 51 tok/s/user on MiniMax M2.5/M2.7 (B200), FP4 delivers 10989 tok/s/chip at $0.04 per million tokens; FP8 delivers 3694 tok/s/chip at $0.13. FP4 is 197% cheaper per token; FP4 delivers 197% more tok/s/chip. Lower-precision quantization trades model accuracy for throughput — check the evaluation page for quality impact.

FP4 posts 6499 tok/s/chip for $0.07 per million tokens at 81 tok/s/user on MiniMax M2.5/M2.7 (B200); FP8 posts 1698 tok/s/chip for $0.28. FP4 is 283% cheaper per token; FP4 delivers 283% more tok/s/chip. Quantization-level accuracy differences are tracked on the evaluation tab.

Throughput at 111 tok/s/user on MiniMax M2.5/M2.7 (B200): FP4 hits 2605 tok/s/chip, FP8 hits 743. Per-million costs land at $0.18 and $0.65 respectively. FP4 is 251% cheaper per token; FP4 delivers 251% more tok/s/chip. The cost-throughput tradeoff from lower precision is only part of the picture — see the evaluation page for accuracy data. (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.)