China's Flash twins clash: MiniMax M3.1 Flash lands two days after DeepSeek's

新智元 · wechat · 2026-10-19

Just two days after DeepSeek's V4.1 Flash-1017, MiniMax released M3.1 Flash, forming a domestic "Flash twin" rivalry. Key numbers:

Hands-on tests included an Inception-style folding-city 3D webpage via Codex and a pagoda generated in one 50K-token API call. The official CLI now defaults to M3.1 Flash; a Pro version targeting higher capability is coming. The piece argues Flash models are taking over tasks once reserved for flagships, with DeepSeek stronger on SWE and MiniMax on enterprise workflows and financial tool use.

Related event: MiniMax Releases M3.1 Flash: 3x Faster, One-Third the Cost of M3(26 posts)→

Original post →

More from coding & agent

coding & agent channel →