Please provide the original title to rewrite.

B.AI reports 1.33 trillion daily tokens after free access to premium AI models drives developer surge

B.AI, an AI infrastructure platform positioning itself at the intersection of Web2 and Web3 settlement, said daily token throughput on its network surpassed 1.33 trillion within days of opening free access to a lineup of “top-tier” AI models. Over a 15-day period, the company reported 8.19 trillion cumulative tokens processed, more than 220,000 new API users, and a total user base exceeding 2.3 million as of September 3.

The spike followed a “zero-cost” rollout designed to remove compute cost barriers for developers. B.AI said it offered unlimited free access to six models: DeepSeek-V4-Flash, DeepSeek-V4-Flash-Vision-Exp, Tencent Hy3, Xiaomi MiMo-V2.5, GLM-5.3-Flash (Ox Alpha), and Qwen3.8-Flash.

On September 3, B.AI introduced a new pricing structure for DeepSeek-V4-Flash and DeepSeek-V4-Flash-Vision-Exp, moving from free access to tiered discounts. The platform said developers receive a 50% discount during peak hours, while off-peak rates drop to 25% of standard peak pricing. It kept zero-cost access in place for GLM-5.3-Flash (Ox Alpha), Qwen3.8-Flash, Tencent Hy3, and Xiaomi MiMo-V2.5, and said overall token throughput continued to rise despite the partial shift toward commercialization.

Why it matters: token throughput is a rough indicator of real usage for model APIs, and B.AI’s figures highlight how aggressively developers respond to lower inference costs. The company framed the milestone as evidence that AI workloads are expanding beyond simple chat interfaces toward more complex, multi-step “agent” workflows that can require frequent switching between models.

B.AI’s broader pitch is to serve as a “global settlement layer for intelligence”—an infrastructure hub intended to coordinate model access, routing, and payment across different providers and agent applications. The company argues that agent-based systems typically rely on multiple models, leaving developers to manage fragmented API standards, billing systems, and rate limits.

To address that, B.AI said it abstracts models into a unified pool of schedulable resources and offers a dual-tier API approach that includes an “official-route” option for reliability and lower-cost custom channels. It also described smart routing on its chat interface designed to send requests to the “optimal” model based on capability and cost.

On the payments side, B.AI said its settlement design supports both traditional payments for Web2-style top-ups and on-chain payment rails for Web3 users, aiming to provide a single integration path for developers operating across different financial and infrastructure setups.

B.AI also highlighted developer tooling aimed at making multi-model workflows easier to run in production. It said it provides native Codex integration with compatibility for the Responses API, allowing developers to use a single B.AI key to run GPT models alongside DeepSeek models within Codex.

The platform described five infrastructure components it is building for agent operations and collaboration:

• x402 Payment Protocol: a “pay-before-response” mechanism intended to support high-frequency, on-chain micro-settlements during cross-agent API calls and compute orchestration.
• 8004 Identity Protocol: on-chain credentials for agents, including execution history and credit scores aimed at establishing trust for collaboration.
• Skills Matrix: standardized tool-calling building blocks designed to interface with MCP servers.
• BAIclaw: a built-in assistant focused on terminal operations and multi-agent workflows.
• BAIcode: a developer-focused assistant intended to streamline task analysis, architecture design, coding, and testing.

In context, the announcement underscores a broader trend in AI infrastructure: as agent-style applications grow more common, platforms are competing to offer cheaper inference, simpler orchestration across multiple model providers, and integrated settlement that works across jurisdictions and payment systems. B.AI’s reported usage surge suggests that pricing and accessibility remain key levers for driving developer adoption, even as platforms begin transitioning parts of their offerings toward paid tiers.