In May 2026, Xiaomi announced MiMo V2.5 Pro pricing that matched the RMB rates quoted for DeepSeek V4 Pro: RMB 3 per million input tokens and RMB 6 per million output tokens. DeepSeek had recently announced a large price cut.
The competition made me wonder how much room providers still had to reduce costs while increasing model size and context length. DeepSeek's low pricing was already well known, but I had not personally tested MiMo, so I could not judge which model performed better.
Price is only one part of the comparison: errors, retries, cache behavior, and how much work the model completes all affect the final cost. These are historical rates from the original post. Consult the MiMo model information and your provider's current price sheet before choosing a plan.
Adapted from the original Chinese article, published on May 27, 2026.
Comments NOTHING