The AI race between the U.S. and China is fragmenting into two distinct philosophies: American labs leveraging massive compute capital, while Chinese developers capitalize on cost efficiency, dominating platforms like OpenRouter. Concurrently, Fastino Labs released GLiNER2.5-Decide, an open-weight 340M parameter model designed for low-latency decision, routing, and classification tasks on consumer CPUs.

We’ll talk about:

• How the AI competition is splitting between high-compute scaling in the U.S. and low-cost, high-efficiency models from China taking up major OpenRouter usage.
• A specialized 340M-parameter open-weight model outperforming larger baselines across structured routing and intent classification tasks.
• Enterprise browser and AI agent security platform securing funding to expand its ...