In a major disruptive move targeting the global developer ecosystem, Xiaomi has officially announced a permanent price reduction of up to 99% for its flagship MiMo-V2.5 series large language model APIs. Effective globally on May 27, 2026, the tech giant is completely restructuring its enterprise AI monetization strategy, matching aggressive market corrections across the generative AI landscape.
Beyond raw dollar savings, Xiaomi is completely discarding traditional pricing models that scale costs upward alongside context length, dealing a direct blow to competitors relying on complex tiered billing.
Breakdown of the New MiMo-V2.5 Tariff Structure
The permanent price reductions heavily favor high-efficiency operations utilizing caching mechanisms, lowering input costs to fractions of a cent per million tokens.
Domestic Pricing (Mainland China)
- mimo-v2.5-pro: Input (Cache Hit) drops to ¥0.025 / million tokens; Input (Cache Miss) falls to ¥3.00 / million tokens; Output pricing sits at ¥6.00 / million tokens.
- mimo-v2.5: Input (Cache Hit) drops to ¥0.02 / million tokens; Input (Cache Miss) hits ¥1.00 / million tokens; Output at ¥2.00 / million tokens.
Global / Overseas Pricing
- mimo-v2.5-pro: Input (Cache Hit) at $0.0036 / million tokens; Input (Cache Miss) at $0.435 / million tokens; Output at $0.87 / million tokens.
- mimo-v2.5: Input (Cache Hit) at $0.0028 / million tokens; Input (Cache Miss) at $0.14 / million tokens; Output at $0.28 / million tokens.
Token Plan Upgrades and Infrastructure Changes
The price restructuring is accompanied by a major overhaul of Xiaomi’s active subscription tiers and developer incentive timelines:
- Context Length Simplification: Developers will no longer face higher base premium costs for scaling inputs within the model’s expansive 1-million token context window.
- Token Plan Allocation Boost: Subscribed enterprise users will see their data limits jump 5× to 8× higher at no extra charge, alongside a complete retrospective reset of all valid account credits used during the billing cycle.
- Incentive Program Conclusion: The massive 100-Trillion Token Creator Incentive Plan, launched in late April, reached full distribution limits ahead of schedule on May 26, prompting Xiaomi to transition entirely to this permanently lowered pay-as-you-go architecture.
Why Big Tech is Pivoting to Near-Zero Token Costs
Industry analysts point to optimization breakthroughs in hardware inference performance and model distillation as the primary catalysts allowing Xiaomi to absorb such dramatic margins. By pushing cache hit costs down to near-zero levels, the firm aims to capture high-volume enterprise pipelines and autonomous agent architectures that continuously loop prompts and generate substantial long-context background tasks.
This strategy forces key global competitors to re-evaluate proprietary pricing walls or risk losing the developer market share built over the previous fiscal year.