
Vercel Agent sessions in the dashboard and Slack now cost 67% less on average. The reduction is already live and does not require users to change their settings, prompts or existing workflows.
According to the Vercel changelog, the team rebuilt how the agent manages context. Each session now needs fewer tokens, while improved prompt caching allows more existing context to be reused across turns.
- Average session cost reduction
- 67%
- across the dashboard and Slack
- Vercel Agent fee
- $0.25 per million tokens
- added to model provider inference rates
How Vercel Agent Uses Fewer Tokens
Vercel has changed four parts of the agent’s operation. Updated workflow-based CLI skills now give it clearer instructions for common tasks and error recovery. Other agents that use those CLI skills also receive the updated guidance.
Context is now managed incrementally as a conversation grows. This includes long Slack threads, where carrying the full conversation into every turn can increase token use. The agent also retains more reusable context in the prompt cache, allowing it to read that material from the cache instead of processing it again.
Coding and planning guidance has also been revised. Vercel says the agent can now plan changes and write code with fewer tokens. Together, these changes reduce the amount of model inference required for an average session rather than changing the price charged for each token.
Billing Rates Stay the Same
The billing formula has not changed. Paid sessions continue to use the model provider’s inference rates, with an additional Vercel Agent fee of $0.25 per million tokens.
The lower average session cost comes from reduced token consumption and better cache reuse. Actual session costs can still vary with the selected model, conversation length and the amount of work completed during that session.
Frequently asked questions
What is Vercel Agent?
Vercel Agent is an agent available through the Vercel dashboard and Slack. It uses coding, planning and CLI workflow guidance to complete tasks and recover from common errors.
How much cheaper are Vercel Agent sessions?
Sessions now cost 67% less on average in the dashboard and Slack. The reduction comes from lower token use and improved prompt-cache reuse, so the cost of an individual session can vary.
How is Vercel Agent billed?
Paid sessions are billed at the chosen model provider’s inference rates. Vercel adds an Agent fee of $0.25 per million tokens, and those billing rates have not changed.
Sources
3 checkedHow we cover tool news: Create With's tool desk drafts these reports with AI from the sources listed above and checks them against those sources before publishing.





