AI Glossary
What is Test-Time Scaling?
Overview
Test-time scaling improves an AI system's answer by spending more computation during inference, for example through longer reasoning, multiple candidate solutions, search, or verification. It matters because models can often solve harder tasks without retraining when the system allocates compute adaptively and uses reliable methods to select or check the result.
Why it matters
Test-time scaling shifts part of model performance from training into the inference system, creating new tradeoffs among accuracy, latency, and cost.
Where it appears in AI research
- Reasoning model evaluations
- Agent planning and search systems
- Process reward and verification research
- Inference cost and latency analysis
Related terms
Related DeepSignal articles
KV-PRM: Efficient Process Reward Modeling via KV-Cache Transfer for
KV-PRM introduces an efficient process reward model that leverages KV cache to reduce scoring costs from O(L^2) to O(L), significantly enhancing multi-agent test-time scaling. It outperforms traditional text-based PRMs across benchmarks like MATH and GSM8K, achieving up to 5,000x reduction in scoring FLOPs and 37x reduction in latency.