On August 24, AI infrastructure platform B.AI announced that cumulative token throughput across its sitewide free-access campaign officially surpassed the 2 trillion threshold. The milestone follows a series of campaigns offering free access to prominent artificial intelligence models.
The momentum began on August 17 when B.AI provided free access to DeepSeek V4 Flash, resulting in a single-day throughput of 220 billion tokens. Subsequent free-access offerings included Tencent Hy3, DeepSeek-V4-Flash-Vision-Exp, and Xiaomi MiMo-V2.5, alongside earlier initiatives featuring MiniMax M3 and Qwen-3.8 MAX.
Dual-Tier API Model and Cost Reductions
To address growing compute demands, B.AI operates as a global compute distribution hub featuring a dual-tier API access model divided into Official and Custom Provider channels:
- Official Channel: Designed for core production environments, this tier features direct API connections and guarantees platform availability. B.AI applies baseline price cuts ranging from 10% to 40% off market rates.
- Custom Provider Channel: Aimed at non-critical workflows, this option routes traffic through third-party vendors such as Mix, Nebula, and OL Station with live discounts reaching up to 90% off standard pricing.
Additionally, the platform incorporates a native Auto mode within its chat interface to analyze user intent and dynamically route natural language queries to optimal underlying models.
Payment Integrations and Global Infrastructure
B.AI supports a unified payment infrastructure that combines traditional fiat rails—including Visa, WeChat Pay, Alipay, and UnionPay—with cryptocurrency settlement networks. Based in Singapore, the platform continues to scale its tiered free access events and subsidized compute offerings for enterprise and developer ecosystems.
