Charged based on No of Requests and Duration (per GB-seconds).
The total cost is calculated by adding the Requests charge and the Duration charge. For example, if your function uses 1GB of memory and runs for 1 second, that would be 1 GB-second. The cost would be No of requests * duration cost (1GB - Second cost)
The default concurrency limit across all functions per AWS region in a given account is 1000/sec.
This 1,000 concurrent execution limit is a shared pool that all of your Lambda functions in that account can use.
If one of your Lambda functions receives 1,000 or more requests, and then another function receives additional requests, the second function’s requests will be rejected due to throttling, as there is no more available concurrency in the shared pool. [Throws 429 error code]
To avoid this throttling issue, you can set a reserved concurrency limit on individual Lambda functions, which reserves a portion of the 1,000 concurrent execution limit for that specific function. AWS Docs