In vLLM v0.4.2 SchedulerConfig, when max_num_batched_tokens is unset and enable_chunked_prefill=False, code sets max_num_batched_tokens to max(max_model_len, 2048).

Setting

Applies when

Evidence

623:                 # If max_model_len is too short, use 2048 as the default value
624:                 # for higher throughput.
625:                 self.max_num_batched_tokens = max(max_model_len, 2048)
626:         if enable_chunked_prefill:

Claim support

No independent claim-support receipt is recorded. Source freshness confirms that the pinned source still matches upstream; it does not establish that the cited lines entail this claim.

Does not establish

• defaults in later vLLM releases
• the value for deployments that explicitly set max_num_batched_tokens

Freshness

Initial fast-decay policy: re-check within three days and treat as stale after seven days without verification.