MiniMaxAI/MiniMax-M3-MXFP8

Quantization vs MiniMaxAI/MiniMax-M3. MiniMaxAI/MiniMax-M3-MXFP8 is a quantized version of the other (F8_E4M3/U8).

3 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse