MiniMax has released M3.1 Flash, a new reasoning model that scores 42 points on the Artificial Analysis Intelligence Index. That score sits 13 points higher than its predecessor M3. The model achieves this increase at less than half the cost per task.
New model competes with major rivals
The M3.1 Flash model ties with GLM-5.3-Flash and edges past Gemini 3.8 Flash, which scores 41. It trails Deepseek V4.1 Flash 1017, which reaches 45 points on the index. Complex reasoning saw the biggest improvement in the new release. On Humanity's Last Exam, M3.1 Flash hits 48 percent. That performance beats both Deepseek and GLM.
M3.1 Flash scores 22 percent on the CritPt benchmark, which tests research-level physics reasoning across 71 complex problems. That puts the model ahead of GLM-5.3-Flash at 15 percent and Gemini 3.8 Flash at 18 percent. It sits just behind Deepseek at 24 percent. For factual accuracy on the AA-Omniscience test, M3.1 Flash answers 56 percent of questions correctly. That result stands well above GLM-5.3-Flash at 28 percent and Deepseek at 43 percent.
Efficiency gains lower task costs
The model uses far fewer output tokens than its rivals. This efficiency brings the cost per task down to about $0.22. MiniMax states the model runs three times faster than M3. One weak spot is its 93 percent hallucination rate. Artificial Analysis notes the model tries to answer every test question instead of declining when it is uncertain.
The model handles up to one million tokens of context and accepts both text and images. API pricing sits at $0.15 per million input tokens and $0.60 per million output tokens. Cached inputs cost $0.015 per million tokens. MiniMax pulls ahead of Deepseek on task cost due to needing fewer output tokens across the benchmark, using about 189 million compared to 224 million for Deepseek.
Open weights release follows API launch
The API is live at platform.minimax.io. MiniMax also offers a desktop app called MiniMax Code. MiniMax plans to release the model as open weights on October 23. Deepseek has already released its competing model on Hugging Face.
