Reports say that DeepSeek V4's API gross margin is as high as 70–80%, which is pretty impressive considering how cheap their pricing is. Their pricing strategy only differentiates between peak and off-peak adjustments. minimax and mimo are likely undervalued. High volume sales might not necessarily earn less than high-quality sales. The minimax M3 model, while honestly not great to use and prone to errors, still ranks high in call volume. With less than 500 billion parameters and 23 billion active parameters, compared to V4's 1.6 trillion parameters and 49 billion active parameters, minimax's active parameters are only half as much. In terms of programming scores, M3 actually outperforms V4, and their pricing is almost identical. Although their optimization might not be as extreme as DeepSeek's, their gross margin is estimated to be around 60–70%. Goldman Sachs mentioned in their report that minimax's annualized revenue growth is expected to accelerate significantly, and this seems absolutely on point. The stock's poor performance is mainly due to chip supply issues, the impact of unlocking shares on the market, and the company's bond financing, which diluted shareholder equity. The company's competitive edge isn't completely lagging behind. I think people might be underestimating minimax—their actual technical capabilities aren't as bad as everyone assumes. Their product positioning focuses on value for money, high volume, and unlimited usage. Achieving this with only 23 billion active parameters is actually quite impressive. The large-parameter M3 Pro model might very well squeeze into the top tier.