Mercury 3 now available on AI Gateway
Vercel’s AI Gateway now offers Mercury 3 from Inception at a 50% launch discount, featuring diffusion-based text generation with over 1,000 tokens per second and a 1M-token context window.
Vercel has integrated Mercury 3, a diffusion-based text generation model developed by Inception, into its AI Gateway platform with a 50% discount on launch pricing. The model delivers generation speeds exceeding 1,000 tokens per second, enabling real-time applications such as browser agents, interactive education, and customer support. Mercury 3 employs parallel refinement to improve reasoning in latency-sensitive workloads, addressing the need for rapid, high-quality responses in dynamic environments.
Mercury 3 introduces a 1 million-token context window, allowing applications to process long documents and extended conversation histories without truncation. This feature is particularly valuable for use cases requiring deep context retention, such as real-time search and extended dialogue systems. The model is accessible through popular coding agents including Claude Code, Codex, and Cursor, simplifying integration for developers.
To deploy Mercury 3, users must install the latest Vercel CLI and configure the model via the agent setup process. The model is selected by specifying inception/mercury-3 in the agent configuration, with additional guidance available in Vercel’s coding agents guide. This streamlined process reduces deployment complexity for teams leveraging AI-driven workflows.
AI Gateway serves as a unified API for model inference, offering features such as usage tracking, cost management, and configurable retries for improved reliability. The platform supports Zero Data Retention, budget controls for API keys, and routing rules to optimize performance. AI Gateway adheres to provider pricing without markup and does not impose platform fees on inference, including Bring Your Own Key requests.