Normalized comparison

Google Cloud Run vs AWS Lambda Limits

Cloud Run services provide longer HTTP request time and container concurrency, while Lambda has distinct invocation payload modes and a 15-minute function ceiling.

Verified Aug 22, 2026Google Cloud RunAWS Lambda
Values are comparable only within the scope shown. Runtime, plan, model, invocation mode, and account-specific capacity can change the effective result.

Quick comparison

CriterionGoogle Cloud RunAWS LambdaComparability note
Maximum service/function timeGoogle Cloud Run60 minutesCloud Run servicesGoogle CloudAWS Lambda900 seconds (15 minutes)A function-level configuration ceiling; the configured timeout may be lower.Amazon Web ServicesCloud Run's value is an HTTP service request timeout; Lambda's is the whole function invocation.
Ordinary HTTP/synchronous payloadGoogle Cloud Run32 MiB request; 32 MiB non-streamed responseHTTP/1 servicesGoogle CloudAWS Lambda6 MB eachAWS uses MB in Lambda docs to mean 1,024 KB.Amazon Web ServicesCloud Run varies by HTTP version and streaming; Lambda has separate async and streamed response modes.

Which is best for your requirement?

Choose Google Cloud Run when

Its documented maximum service/function time scope fits your measured workload and the caveats shown in the comparison.

Choose AWS Lambda when

Its documented execution or account model better matches the exact requirement rather than a vendor-wide headline.

Validate before migration

Test payload, duration, throughput, concurrency, failure behavior, billing scope, and recovery with representative traffic.

Validate the decision with your workload

Before choosing between Google Cloud Run and AWS Lambda, reproduce the comparison with the exact plans, models, regions, runtimes, and invocation paths you intend to operate. The reviewed rows cover maximum service/function time and ordinary http/synchronous payload; they do not turn different pricing, reliability, developer experience, or ecosystem tradeoffs into one universal score.

  1. Capture representative request sizes, token usage, duration, concurrency, storage, and failure behavior at realistic percentiles.
  2. Test the boundary and the recovery path on both candidates, including throttling, timeouts, partial failure, retries, and cost controls.
  3. Record which scoped observation drove the choice and recheck its official source before migration or a major traffic increase.

Official sources

Related

Inspect the underlying limits