DeepSeek and Xiaomi have made significant price cuts to their AI API offerings, contrasting with price increases from Western labs. DeepSeek made its 75% discount on V4-Pro permanent on May 22, locking output at $0.87 per million tokens. Xiaomi slashed MiMo-V2.5 prices by up to 99% on May 26, with cached input now at $0.0036 per million tokens for the Pro model. The reductions are driven by technical optimizations in inference infrastructure, including hierarchical KV cache optimization that reduces storage and processing costs by around 80%. Meanwhile, OpenAI's GPT-5.5 doubled output prices to $30 per million tokens, and Anthropic's Claude Opus 4.7 shipped with a new tokenizer that can inflate costs by up to 35%. Chinese models now cost 15-30x less than American counterparts on average, with further gaps for cached workloads. This pricing disparity is reshaping the competitive landscape for AI-powered products.

Key facts

• DeepSeek V4-Pro permanent discount: $0.435 input, $0.87 output per million tokens.
• Xiaomi MiMo-V2.5 cached input drops to $0.0036 per million tokens, up to 99% cut.
• OpenAI GPT-5.5 doubled output price to $30 per million tokens.
• Anthropic Claude Opus 4.7 new tokenizer can increase costs by up to 35%.
• Chinese models cost 15-30x less than American counterparts on average.