Vercel has integrated two new specialized AI models into its AI Gateway platform, while also introducing a feature that automatically sorts service providers based on cost and performance. This is part of Vercel's efforts to offer more flexible options for developers building next-generation AI applications.
Key Developments
Specifically, according to Vercel's announcement, xAI's experimental Grok Build 0.1 model is now available under the identifier xai/grok-build-0.1. This model is optimized for agentic coding tasks and does not allow reasoning level customization. Alongside it, Google's Gemini 3.5 Flash is also added to the system, enhancing parallel execution performance and logical reasoning.
In addition to adding models, Vercel also introduced a feature to sort providers based on practical metrics such as cost, Time-to-First-Token (TTFT) latency, or throughput (TPS). The system automatically calculates the priority order at request time, helping engineers optimize resources without modifying source code when price or performance fluctuations occur on the partner side.
Why It Matters
For the AI development community in Vietnam, these tools help reduce the barriers of managing multi-model infrastructure. The ability to automatically filter providers by data security standards combined with sorting by lowest cost will significantly help projects optimize operational budgets when deploying products at scale.