Morgan Stanley: MiniMax-W (00100) M3.1 Flash pricing is competitive; maintains "Overweight" rating with a target price of HK$900

date
11:45 22/10/2026
avatar
GMT Eight
Morgan Stanley noted that MiniMax's newly launched M3.1 Flash model shows significant improvements in agentic workflows, coding, computer use, and cybersecurity, with inference speed nearly 3 times faster and agent task costs reduced by 70%, and is expected to expand enterprise adoption through advantages in higher intelligence, throughput, and cost.
Morgan Stanley issued a research report maintaining an "Overweight" rating on MiniMax-W (00100) with a target price of HK$900. The bank noted that MiniMax officially launched the M3.1 Flash model on Monday (October 19) of this week, and has not yet disclosed model parameters or licensing information. The model shows significant improvements in agentic workflows, coding, computer use, and cybersecurity, with a DeepSWE score of 73.7%, an OSWorld V2 score of 61.7, and a CyberGym score of 82%, among which its OSWorld V2 and CyberGym scores rank at the top among leading Chinese models. The bank noted that although M3.1 Flash is not the lowest-priced, it is competitive among major Chinese Flash large language models, and its agentic task costs are also advantageous. By optimizing the sparse attention mechanism, model inference speed has increased nearly threefold and agentic task costs have been reduced by 70%; in terms of API pricing, input and output tokens are priced at RMB 1 and RMB 4 per million tokens, respectively. M3.1 Flash ranks second among major Chinese Flash large language models on the Artificial Analysis Intelligence Index, behind only DeepSeek V4.1 Flash Oct. The bank believes that the combination of higher intelligence, higher throughput, and lower cost should help expand enterprise adoption. The bank expects M3 Pro to become the flagship large language model demonstrating MiniMax's pursuit of frontier intelligence capabilities, and anticipates its launch by the end of October.