Zubnet AILearnWiki › Test-Time Compute
Fundamentals

Test-Time Compute

Inference-Time Compute, Chain of Thought, Thinking Tokens
Using additional computation during inference (when the model is generating a response) to improve answer quality. Instead of generating an answer immediately, the model "thinks" longer — generating reasoning tokens, exploring multiple approaches, or verifying its own output. More compute at test time produces better answers, especially for complex reasoning tasks.

Why it matters

Test-time compute is the latest scaling paradigm. The first era scaled training compute (bigger models, more data). The current era also scales inference compute (more thinking per question). Models like o1 and Claude with extended thinking show that letting a model reason for 30 seconds often outperforms a model that answers in 2 seconds, even if the fast model is technically larger. This changes the economics: quality becomes a function of how much you're willing to spend per query.

Deep Dive

The simplest form of test-time compute is chain-of-thought: the model generates reasoning steps before the final answer. More sophisticated approaches include: tree-of-thought (exploring multiple reasoning paths and selecting the best), self-consistency (generating multiple answers and voting), and iterative refinement (the model critiques and revises its own output). Each approach uses more tokens (= more compute = more cost) but produces better results.

Extended Thinking

Models like o1 (OpenAI) and Claude with extended thinking generate internal reasoning tokens that the user doesn't see. These "thinking tokens" let the model decompose complex problems, check its work, consider edge cases, and revise its approach — all before producing the visible response. The cost is higher (you pay for thinking tokens) and latency is longer, but accuracy on math, coding, and reasoning tasks improves dramatically.

Scaling Laws for Inference

Research suggests that test-time compute follows its own scaling laws: doubling inference compute (thinking time) produces predictable improvements in accuracy, analogous to how doubling training compute improves pre-training loss. This means you can choose your quality-cost trade-off per query: simple questions get fast, cheap answers; complex questions get longer, more expensive reasoning. This dynamic allocation is more efficient than using the same compute for every query.

Related Concepts

← All Terms
← Tencent Text-to-Speech →