B.AI says its platform has reached more than 1.51 trillion tokens in daily throughput, roughly 71,500 times its level at launch about five months earlier. The company is positioning the service as infrastructure for both large-language-model distribution and automated transactions between AI agents.
The reported growth comes as demand increases for lower-cost, high-volume inference. B.AI aggregates models from providers including GPT and Claude, alongside Chinese systems such as DeepSeek, Qwen, GLM, Kimi and MiniMax. The platform says it has partnerships with Tencent, Alibaba, ByteDance, DeepSeek, Zhipu AI, Kimi and MiniMax.
A distribution layer for Chinese AI models
B.AI is targeting international developers seeking access to Chinese large language models, which have gained adoption through comparatively low API prices and competitive performance. The company identifies DeepSeek-V4-Flash, Qwen3.8-Flash and GLM-5.3 Flash among the most-used models on its platform.
To encourage usage, B.AI offers tiered API pricing, discounts of up to 90%, limited-time free access and other developer incentives. YZi Labs, described as an ecosystem partner, is also helping bring early-stage projects and developer communities onto the platform, according to B.AI.
Settlement infrastructure for AI agents
Rather than operating only as a model-routing service, B.AI is developing a settlement layer intended to connect compute allocation, agent activity and payment. Its architecture is designed to support both on-chain transactions and conventional payment methods, including WeChat Pay, Alipay, UnionPay and Visa.
The company has built the Web3 component around TRON, using the network’s USDT ecosystem as a settlement rail for cross-border compute payments and agent-to-agent micropayments. B.AI cited a TRC-20 USDT supply of more than 94.2 billion as of September 10.
B.AI also says its wider agent stack includes compute scheduling, execution tools, agent wallets and built-in assistants called BAIclaw and BAI code. At the protocol level, the platform is integrating the x402 payment protocol and the 8004 identity authentication protocol to support identity verification and automated settlement between agents.
The 1.51 trillion-token figure and the associated growth claims were provided by B.AI. The company’s stated strategy is to combine model distribution, compute routing and payment infrastructure as AI agents take on more autonomous digital tasks.
