You can now add a throttle parameter to functions, which limits how often function runs can start within a given time period. When the rate limit is hit, runs will be enqueued in FIFO order until capacity becomes available. This is ideal for rate limiting, smoothing out spikes in traffic, and shaping QoS.