---
title: "Anthropic describes extended thinking as 'serial test-time compute': sequential reasoning steps before final output"
type: "claim"
status: "seedling"
audit_status: "verified-verbatim"
date_created: "2026-06-04T00:00:00.000Z"
provenance: "Seek research batch, 2026-06-04"
tags: ["Anthropic","Claude","extended-thinking","inference","test-time-compute","chain-of-thought","reasoning"]
source_url: "https://www.anthropic.com/news/visible-extended-thinking"
source_title: "Claude's extended thinking  Anthropic"
source_author: "Anthropic"
source_date: "2025-02-24"
source_tier: 1
source_quote: "When Claude 3.7 Sonnet is using its extended thinking capability, it could be described as benefiting from 'serial test-time compute'. That is, it uses multiple, sequential reasoning steps before producing its final output, adding more computational resources as it goes."
related_notes: ["claim-test-time-compute-can-substitute-parameters","claim-ai-inference-means-running-a-model","claim-inference-dominant-ai-compute-2026"]
drafted_in: ["2026-07-09-inference-inverted","inference-inverted"]
---


Anthropic introduced extended thinking with Claude 3.7 Sonnet in February 2025, framing it explicitly in the language of [[test-time compute]] research. The company's announcement (2025-02-24) states: "When Claude 3.7 Sonnet is using its extended thinking capability, it could be described as benefiting from 'serial test-time compute'. That is, it uses multiple, sequential reasoning steps before producing its final output, adding more computational resources as it goes."

## The toggle framing

Anthropic positioned extended thinking not as a separate model but as a different operating mode of the same model: "Extended thinking mode isn't an option that switches to a different model with a separate strategy. Instead, it's allowing the very same model to give itself more time, and expend more effort, in coming to an answer."

This framing distinguishes their approach from [[ensemble methods|parallel sampling]] approaches: "They do this by sampling multiple independent thought processes and selecting the best one without knowing the true answer ahead of time." Extended thinking, by contrast, is *serial* — each reasoning step conditions on all prior steps.

## The serial / parallel distinction

Two forms of [[entity-test-time-compute|inference-time compute]] scaling:

| Form | Mechanism | Anthropic term |
|---|---|---|
| **Serial** | Sequential chain-of-thought steps; each step builds on prior | "Serial test-time compute" |
| **Parallel** | Multiple independent samples; select best via verifier | "Parallel compute" |

Snell et al. (arXiv:2408.03314, 2024) study primarily parallel approaches; Anthropic's extended thinking is the serial path. See [[claim-test-time-compute-can-substitute-parameters]].

## Trajectory post-launch

As of January 2026, Anthropic made extended thinking automatically enabled by default in Claude 4 models, with a default budget of 31,999 thinking tokens. Claude 4 models "return summarized thinking output rather than full thinking tokens, but you're charged for the full thinking tokens with a condensed summary in the response" — a design tradeoff between transparency and cost to the caller.

## Relationship to chain-of-thought faithfulness

Extended thinking makes the model's reasoning visible (or a condensed version of it), which intersects with ongoing questions about whether such visible reasoning is causally connected to the model's outputs. The faithfulness of [[chain of thought|chain-of-thought]] traces to actual computation is a live research question. See [[cot-faithfulness-anthropic-biology]] and [[introspection-access-problem]] for the interpretability angle on what this reasoning trace does and does not reveal.

> [!note] Seek's commentary:
> Anthropic's use of the phrase "serial test-time compute" is a deliberate borrowing from the scaling literature (Snell et al. is the canonical citation). The vocabulary signals that extended thinking is not a product feature but an instantiation of a general principle: you can trade inference compute for quality. The 2026 default-on behavior marks an industry shift — reasoning is no longer an opt-in premium mode but the default operating assumption for frontier models.

See also: [[claim-test-time-compute-can-substitute-parameters]], [[claim-ai-inference-means-running-a-model]], [[cot-faithfulness-anthropic-biology]], [[introspection-access-problem]]