ModelRefs / Serverless Inference — AI Glossary

Serverless Inference — AI Glossary

A deployment model where inference capacity scales to zero when idle and scales up automatically per request, billed per-token. Also called on-demand inference.

Overview

Serverless LLM inference (AWS Bedrock, Together AI, Modal, Replicate) eliminates idle GPU cost by spinning down between requests. Cold starts (loading weights) add 1–10 s latency. Ideal for bursty or low-traffic workloads. Contrast with dedicated endpoints (always-on GPU, predictable latency, higher baseline cost).

Reference details

Topicinference
Also known ason-demand inference
Last reviewed2026-06-24

Example: Where the break-even sits

A dedicated endpoint bills continuously whether or not traffic arrives; serverless bills per token but adds a cold start of seconds when scaled to zero. For bursty or low-volume workloads serverless is far cheaper. For steady high traffic the dedicated endpoint wins on both cost and latency — the crossover is a traffic-shape question, not a preference.

Commonly confused with

Serverless does not mean no cold start. Scaling to zero is precisely what creates one: multi-gigabyte weights have to be loaded before the first token. Providers mitigate it with warm pools, but the trade-off is inherent to scaling to zero.

When to use it

Reach for it when:

  • Traffic is bursty, unpredictable or low-volume
  • Idle GPU cost is the dominant line item
  • A first-request delay is acceptable

Reach for something else when:

  • Interactive products where a cold start lands on a real user
  • Steady high throughput, where dedicated capacity is cheaper per token
  • Strict tail-latency SLOs

Continue your research

Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Serverless Inference — AI Glossary.

Frequently asked questions

What is Serverless Inference?

A deployment model where inference capacity scales to zero when idle and scales up automatically per request, billed per-token.

Is Serverless Inference the same as on-demand inference?

Yes — on-demand inference are common aliases for Serverless Inference.

What concepts are related to Serverless Inference?

Closely related concepts include cold start, warm pool, inference provider.