B300: FP4 vs FP8 Precision Comparison

At 122 tok/s/user on MiniMax M3 428B (B300), FP4 delivers 5953 tok/s/chip at $0.11 per million tokens; FP8 delivers 3849 tok/s/chip at $0.16. FP4 is 55% cheaper per token; FP4 delivers 55% more tok/s/chip. Lower-precision quantization trades model accuracy for throughput — check the evaluation page for quality impact.

FP4 posts 3098 tok/s/chip for $0.20 per million tokens at 235 tok/s/user on MiniMax M3 428B (B300); FP8 posts 2237 tok/s/chip for $0.28. FP4 is 39% cheaper per token; FP4 delivers 39% more tok/s/chip. Quantization-level accuracy differences are tracked on the evaluation tab.

Throughput at 349 tok/s/user on MiniMax M3 428B (B300): FP4 hits 2067 tok/s/chip, FP8 hits 1040. Per-million costs land at $0.30 and $0.60 respectively. FP4 is 99% cheaper per token; FP4 delivers 99% 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 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.)