MiniMax M3.1 Flash: half the price of M3, but not faster or better

MiniMax released M3.1 Flash on October 19, 2026, priced at $0.15 per million input tokens and $0.60 per million output tokens — half of M3's rates. In a three-task agent test, Flash matched M3's accuracy but ran slower, and the claimed 3x inference speedup did not hold.

• Flash pricing: $0.15/1M input and $0.60/1M output tokens vs $0.30 and $1.20 for M3
• Artificial Analysis scored M3.1 Flash at 72.5% on AutomationBench-AA
• Both models scored 10/10 on the incident test, but Flash cost about 68% less
• Neither model answered the ~500K-token long-context question on the first run