B.AI hits 1.33 trillion daily tokens after free model access
B.AI, an AI infrastructure platform, reported a rapid usage surge after opening free access to multiple premium models. Daily token throughput surpassed 1.33 trillion. Over 15 days, cumulative volume reached 8.19 trillion tokens and more than 220,000 new API users joined. As of September 3, the total user base exceeded 2.3 million. The rollout is part of a broader strategy to build a global settlement layer for intelligence that supports cross-node collaboration, orchestration, and value distribution for AI agents.
Usage Growth and Pricing Structure
B.AI’s zero-cost model lineup removed cost barriers and expanded access. The platform offers 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 tiered discounts for DeepSeek-V4-Flash and DeepSeek-V4-Flash-Vision-Exp: 50 percent discounts during peak hours and off-peak pricing at 25 percent of peak. Zero-cost access remains for GLM-5.3-Flash (Ox Alpha), Qwen3.8-Flash, Tencent Hy3, and Xiaomi MiMo-V2.5. Token throughput continued to climb following the pricing update.
Global Settlement Layer and “AI Grid”
B.AI positions itself above models and below agents, unifying diverse providers, capabilities, and costs into a schedulable resource pool. A dual-tier API offers official-route connections alongside lower-cost custom channels. Smart routing in the Chat interface directs requests to models that meet performance and cost targets. B.AI describes this as an “AI grid” designed to support autonomous agent workflows at scale.
The settlement layer supports Web2 and Web3 payments. Traditional methods enable low-friction top-ups. On-chain rails offer decentralized, verifiable, and low-friction options. Developers can select settlement paths that match their infrastructure with single-point integration for global reach.
Execution Layer and Developer Integration
B.AI integrates natively with Codex and supports the Responses API, allowing a single B.AI key to run GPT models and DeepSeek models in the same workflow. This enables code generation, reasoning, debugging, and refactoring with unified tooling.
Five Core Components for Agent Operations
- x402 Payment Protocol: Pay-before-response model executing high-frequency on-chain micro-settlements during cross-agent API calls and compute orchestration.
- 8004 Identity Protocol: On-chain credentials for agents with execution history and credit scores to support cross-node trust.
- Skills Matrix: Standardized, plug-and-play tools interfacing with MCP servers for modular tool-calling.
- BAIclaw: Assistant for terminal operations and multi-agent workflows.
- BAIcode: Developer assistant covering task analysis, architecture, coding, and testing.
Key Takeaways
- Daily token throughput exceeded 1.33 trillion; 8.19 trillion tokens over 15 days; 220,000 new API users; 2.3 million total users by September 3.
- Unlimited free access to six models remains; tiered discounts applied to DeepSeek-V4-Flash and DeepSeek-V4-Flash-Vision-Exp.
- Dual-tier API and smart routing form an “AI grid” that abstracts models and optimizes for cost and reliability.
- Supports Web2 and Web3 settlement with single-point integration.
- Execution stack includes Codex integration and five core components: x402, 8004, Skills Matrix, BAIclaw, and BAIcode.
B.AI reports that its free-access campaign accelerated usage and supports a strategy focused on unified model orchestration, decentralized settlement options, and agent-ready infrastructure.
Authors credited by the source: BlockmanPR
Published by Holder based on an external source.
