Vercel Agent sessions now cost 67% less
Vercel reduced Agent session costs by 67% through token efficiency improvements in context management and prompt caching, now live without user action.
Vercel announced a 67% average reduction in Agent session costs for dashboard and Slack interactions, achieved by rebuilding how context is managed. The update decreases tokens required per session and enhances prompt-cache reuse, lowering computational overhead. No user action is required as the changes are already deployed across supported platforms. The adjustments apply to both new and existing sessions.
Updated Vercel CLI skills now provide agents with clearer instructions for common workflows and error recovery, improving reliability. These improvements also benefit other agents utilizing the same skills. The updates focus on standardizing responses and reducing ambiguity during automated tasks. The changes are designed to streamline repetitive workflows without additional configuration.
Agent’s context management has been optimized to handle growing conversations more efficiently, including long Slack threads. The system now preserves reusable context across interactions, allowing more data to be retrieved from cache rather than reprocessed. This reduces token consumption during extended exchanges. The feature is active by default for all users.
Vercel Agent’s billing structure remains unchanged, with paid sessions billed at model provider rates plus a $0.25 per million tokens fee. Users can monitor usage and costs in the Agent section of the Vercel dashboard. The updates aim to improve cost predictability while maintaining performance. No changes to pricing tiers or subscription plans are required.