MiniMax officially launches M3.1 Flash at $0.15 per million input tokens

MiniMax formally rolled out M3.1 Flash on October 19, 2026, a day after hinting that the anonymous Space Bunny model was its work. The company's launch post confirms that Space Bunny was an early build and says it handled more than 200 trillion tokens before anyone knew who made it.

MiniMax pitches the model at agent work, coding, computer use and security. Its own numbers include 76.2 on BankerToolBench, 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2 and 82% on CyberGym. On OSWorld V2 and Agents' Last Exam it claims first place among Chinese models. It says inference is close to three times faster than before, with output often above 200 tokens per second, and that agent tasks cost about 70% less than on the previous model.

Pricing is flat: $0.15 per million input tokens, $0.60 per million output tokens and $0.015 for cached input. The model is live in the MiniMax API, the M Plan subscription and the MiniMax Code desktop app, with weights promised later and a larger Pro model also on the way. Independent tester Artificial Analysis scored it 42 on its Intelligence Index, 13 points above M3, but found it often gives wrong answers instead of admitting uncertainty, VentureBeat reported.

Why it matters

Cheap Chinese models keep closing the gap on agent tasks, which pushes down what teams pay to run coding and automation agents, though the reported hallucination rate means outputs still need checking.

Source: MiniMax