Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Reasoning model

A reasoning model is a language model trained to work through a problem in a stream of intermediate thinking tokens before giving its final answer, which tends to help on maths, coding and multi-step tasks.

Also called: thinking model, extended thinking

By Divya Sharma · Checked Nov 15, 2026 · How we review

Reasoning models spend extra compute at answer time, a pattern called test-time compute. Instead of replying straight away, the model writes out a working process, checks steps and changes course, then answers.

How they are trained

The behaviour grows out of chain-of-thought prompting, but reasoning models are trained for it, often with reinforcement learning on tasks whose answers can be checked. DeepSeek's R1 paper reports that reasoning abilities can be incentivised through pure reinforcement learning, with patterns such as self-reflection and verification emerging during training.

Extended thinking

Providers expose reasoning in different ways. Anthropic calls its version extended thinking. In its manual mode, a budget_tokens setting gives a target for how many tokens Claude may spend on its internal reasoning, and Anthropic's documentation recommends its newer adaptive thinking mode, where the model decides how much to think.

Costs and limits

Thinking tokens are usually billed as output tokens, so a reasoning model can cost much more per request than a plain one and take longer to reply. See token pricing. For simple lookups, rewrites or classification, a non-reasoning model is often faster and just as accurate. The visible reasoning is not a guaranteed account of how the model reached its answer.

Related terms

Chain-of-thought
Chain-of-thought (CoT) is a prompting method in which a language model writes out intermediate reasoning steps before its final answer, which tends to improve results on maths, logic and other multi-step problems.
Large language model
A large language model (LLM) is a neural network with billions of parameters, trained on large amounts of text to predict the next token, which lets it write, answer questions, summarise and follow instructions.
Token pricing
Token pricing is how most AI model APIs charge: a set price per million input tokens you send and a separate, usually higher, price per million output tokens the model generates, with discounts for cached input and batch jobs.
Test-time compute
Test-time compute is the amount of computation a model spends while answering a request, rather than during training. Letting a model think longer or try several answers can improve results on hard problems.
Output tokens
Output tokens are the tokens a language model generates in its reply. They are counted and billed separately from input tokens, and a max_tokens setting caps how many one response may contain.

Sources