Fluid compute
A serverless execution model proposed by Vercel that allows multiple concurrent requests to share a single instance, retains an initialized runtime environment between requests, reduces cold-start overhead, and bills by actual CPU compute time.
Timeline (last 90 days)
Vercel's compute forms are divided into two lines: request-driven applications scale elastically through Fluid compute, while long-running pull-based workloads are handled by Vercel Daemons.
Fluid compute allows instances to remain alive to some extent between requests to reduce cold start frequency, but does not guarantee long-term instance residency.
Fluid compute is a compute model Vercel previously introduced that adds instance reuse and concurrent processing on top of traditional Serverless, relaxing the scheduling granularity from "one request per instance" to "one instance, multiple tasks."
Vercel containers run on Fluid compute, scale with traffic, and use Active CPU pricing, so idle time spent waiting for API responses, database queries, or LLM inference is not billed.
Vercel Daemons has entered public beta, allowing containers to be deployed as continuously running background worker processes on Fluid compute, with support for specifying the number of instances per region and automatically maintaining capacity and restarting crashed processes.
Fluid compute lets multiple concurrent requests share the same instance and preserves the initialized runtime environment between requests, thereby reducing cold start overhead and improving resource utilization.
All Facts (6)
Vercel Daemons has entered public beta, allowing containers to be deployed as continuously running background worker processes on Fluid compute, with support for specifying the number of instances per region and automatically maintaining capacity and restarting crashed processes.
60%UnverifiedVercel's compute forms are divided into two lines: request-driven applications scale elastically through Fluid compute, while long-running pull-based workloads are handled by Vercel Daemons.
50%UnverifiedFluid compute allows instances to remain alive to some extent between requests to reduce cold start frequency, but does not guarantee long-term instance residency.
50%UnverifiedFluid compute is a compute model Vercel previously introduced that adds instance reuse and concurrent processing on top of traditional Serverless, relaxing the scheduling granularity from "one request per instance" to "one instance, multiple tasks."
50%UnverifiedVercel containers run on Fluid compute, scale with traffic, and use Active CPU pricing, so idle time spent waiting for API responses, database queries, or LLM inference is not billed.
50%UnverifiedFluid compute lets multiple concurrent requests share the same instance and preserves the initialized runtime environment between requests, thereby reducing cold start overhead and improving resource utilization.
50%