interleaved thinking
In plain wordsA way for an AI agent to pause and reason again after every action it takes — reading each search result or test output before deciding what to do next — instead of only thinking once at the start.
A mode of reasoning modelreasoning model, n.: An AI model trained to work through a problem step by step in hidden 'thinking' text before it answers. Letting it think longer generally makes it better at math, coding and planning. operation in which the model produces fresh reasoningreasoning trace, n.: The written-out 'thinking' an AI reasoning model produces before its final answer — notes to itself full of attempts, checks and corrections, which users may or may not get to see. between tool callstool call, n.: A single request from an AI model to use one of its tools — naming the tool and filling in the details — which the surrounding software then carries out and reports back on. within a single assistant turn, reflecting on each tool result before choosing its next action, rather than reasoning only before its first output. It matters for agentsAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done. running an agent loopagent loop, n.: The repeating cycle at the heart of an AI agent: the model decides what to do next, software carries it out, the result is shown to the model, and the cycle repeats until the task is finished., where the right next step depends on what the last step returned. Supporting it requires the harnessharness, n.: The software around an AI model that gives it instructions and tools, carries out its actions and decides when it stops. The model is the engine; the harness is the rest of the car. to pass earlier thinking blocks back to the model unchanged, so that its reasoning stays continuous across the turn. The idea carries the ReActReAct, prop. n.: A method in which an AI alternates between thinking out loud about what to do next and taking an action, such as a web search, then uses what it observed to plan its next step. pattern of alternating thoughts and actions over to trained reasoning models.
With interleaved thinking the agent noticed the failing test and changed its plan mid-turn.
Origin
Anthropic introduced interleaved thinking as a beta feature for its Claude 4 models in May 2025; from the Claude 4.6 generation it became automatic under adaptive thinking. Agentic reasoning models from other developers adopted similar behavior during 2025.
Related
- extended thinkingextended thinking, n.: A setting in some AI models, notably Anthropic's Claude, that lets the model write out private step-by-step reasoning before its final answer, improving results on hard problems.
- tool usetool use, n.: An AI model's ability to reach beyond its own knowledge by operating outside tools — searching the web, running code, using a calculator or an app — and folding the results into its work.
- agent loopagent loop, n.: The repeating cycle at the heart of an AI agent: the model decides what to do next, software carries it out, the result is shown to the model, and the cycle repeats until the task is finished.
- ReActReAct, prop. n.: A method in which an AI alternates between thinking out loud about what to do next and taking an action, such as a web search, then uses what it observed to plan its next step.
- reasoning modelreasoning model, n.: An AI model trained to work through a problem step by step in hidden 'thinking' text before it answers. Letting it think longer generally makes it better at math, coding and planning.
- harnessharness, n.: The software around an AI model that gives it instructions and tools, carries out its actions and decides when it stops. The model is the engine; the harness is the rest of the car.
Sources
- Anthropic (2026). Extended thinking. Claude Developer Platform documentation. platform.claude.com/docs/en/build-with-claude/extended-thinking
Mentioned in
extended thinkingextended thinking, n.: A setting in some AI models, notably Anthropic's Claude, that lets the model write out private step-by-step reasoning before its final answer, improving results on hard problems.