prompt

A hidden cost-saver in AI services: reusing the computation for repeated prompt prefixes (system instructions, documents) instead of recalculating every time.

What's it called?

Q: What's it called?

Prompt caching. Prompt caching stores and reuses computation for repeated inputs, slashing cost and latency — especially vital for agents that re-read the same context dozens of times.

Prompt caching ✓
Prompt laundering
Prompt pickling

Prompt caching
Prompt caching stores and reuses computation for repeated inputs, slashing cost and latency — especially vital for agents that re-read the same context dozens of times.

▶ Quration AIQ

prompt

Designing instructions to get what you want became a skill — briefly hyped as a 'six-figure job.' What's it called?What is this technique of including examples in your prompt called?What is this technique of making the model write out its reasoning first?What's the acronym for this 'retrieval-augmented generation'?What is this called?What's it called?What is this numerical representation called?What is this emerging skill called?What's the common name for this feature that retains user information across sessions?What made the leap possible?What does ChatGPT call these custom chatbots?What's this technique of assigning the AI a role to shape its perspective and depth?

Quration — Quration AIQ