概述
More thinking often improves results on hard problems like math, coding and multi-step analysis, but it adds cost and delay, and past a point it stops helping. Choosing the right level per task is one of the simplest ways to balance quality, speed and price.
深入探讨
Reasoning models are trained to produce a chain of intermediate reasoning before their final answer. That reasoning is made of tokens, just like the answer, and generating it takes time and money. Providers let you control how much of it happens. OpenAI exposes a reasoning_effort parameter on its reasoning models with levels such as low, medium and high; some newer models add a minimal level. Anthropic's extended thinking on supported Claude models takes a budget_tokens value that sets the maximum number of tokens Claude may use for thinking; the budget has a minimum of 1,024 tokens and must be smaller than max_tokens. Google's Gemini 2.5 models accept a thinking budget, and some can have thinking turned down or off. The names differ, but the idea is the same: a dial between fast and thorough. These settings are guidance, not exact counts. A model given a large budget may use only part of it on an easy question, and effort levels shift behavior rather than fixing a number. Cost is the part people often miss. Reasoning tokens are generally billed as output tokens even when you cannot see them, and output tokens usually cost more than input tokens. A short visible answer can carry a large hidden bill. Latency grows too, since every reasoning token is generated in sequence. More thinking has diminishing returns. On easy or factual lookup questions, extra reasoning rarely improves accuracy and can sometimes lead a model to overcomplicate or second-guess a correct answer. Gains are largest on problems that genuinely need several steps. A common misconception is that the highest setting is always best. The practical approach is to test your own tasks at several levels and pick the lowest one that meets your quality bar.
战略影响
速度与规模
语言工作流程可以在不牺牲一致性的情况下更快地移动。
交通与覆盖范围
它扩展了跨语言和沟通方式的访问。
更清晰的判决
团队可以花更多时间进行判断，而自动化则可以处理重复。
The Future of Reasoning Effort and Thinking Budgets
Providers are moving toward models that decide for themselves how long to think, with effort settings acting as bounds rather than fixed instructions; Anthropic's newer models, for example, offer adaptive thinking, and some products already switch automatically between fast and reasoning modes. Research continues into making reasoning more efficient so that similar accuracy needs fewer tokens. Parameter names, allowed values and pricing have changed several times and will likely keep changing, so treat current settings as provider-specific details and re-run your own evaluations when you upgrade models.
现实世界的实施
A customer support bot answering simple order questions uses low reasoning effort, which keeps replies fast and cheap without hurting accuracy.
A developer debugging a tricky concurrency bug switches to high effort for that one request and gets a correct diagnosis the low setting missed.
A team using Anthropic's extended thinking sets a budget of a few thousand tokens for routine document review and a much larger budget only for contract clauses flagged as unusual.
An analyst runs the same set of 50 test questions at low, medium and high effort and finds medium matches high on accuracy for their tasks at a lower cost, so they standardize on medium.
风险与防护栏
幻觉的事实可以悄悄地进入报告、支持流程或研究成果。
及时的敏感性可能会在类似的请求中产生不一致的结果。
如果访问控制薄弱，敏感文本数据可能会暴露。
实施路线图
在推出之前定义输出格式、语气和质量标准。
当准确性很重要时，请使用可信来源进行地面响应。
为高风险输出保留人工审查检查点。
跟踪故障模式并定期重新训练提示或工作流程。
不断探索
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Reasoning Effort and Thinking Budgets quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
常见问题
What is Reasoning Effort and Thinking Budgets?
Reasoning effort and thinking budgets are settings on reasoning models that control how much internal step-by-step work the model does before answering. More thinking often improves results on hard problems like math, coding and multi-step analysis, but it adds cost and delay, and past a point it stops helping. Choosing the right level per task is one of the simplest ways to balance quality, speed and price.
What do reasoning effort and thinking budget settings control?
These settings adjust how much intermediate reasoning a reasoning model generates before its final answer.
How are hidden reasoning tokens generally billed?
Reasoning tokens are generally billed as output tokens, which usually cost more than input, even when they are not shown.
For Anthropic extended thinking, which rule about budget_tokens is stated in the guide?
The budget has a minimum of 1,024 tokens and must be below max_tokens so there is room for the answer.
A model is given a large thinking budget for an easy question. What typically happens?
Budgets are upper limits and guidance. On easy questions the model often uses much less.
Where do larger reasoning settings help the most?
Gains are largest on multi-step problems. Easy tasks rarely benefit and can even suffer from overthinking.
继续学习
相关指南
为此主题精选的更多指南