ChinaMiniMax’s M3.1 Flash charges ¥4 per million output tokens.

DeepSeek’s V4.1 Flash, released 17 October, charges ¥8 at peak hours.

MiniMax’s own benchmarks put it level with DeepSeek’s on agent tasks, and it generates 200+ tokens a second.

On the DeepSWE coding test, DeepSeek’s model still leads, 77.0 to 73.7.

MiniMax’s Pro model is the next release developers are waiting for.

How each outlet framed it
Zhidx leans favorable
compares MiniMax M3.1 Flash to DeepSeek V4.1 Flash on yuan-per-token pricing and benchmark gaps, framing China's Flash tier as a 'dual-strong' race

Sources: Zhidx