---
title: "Self-Consistency Improves Chain of Thought Reasoning in Language Models"
description: "A model can reach the same answer along several different routes. This paper shows how sampling those routes and voting on their final answers improves reasoning accuracy without additional training, then exposes the extra computation and the gap between agreement and correctness."
canonical_url: "https://fanout.sh/daily/2026-10-05-self-consistency-reasoning"
md_url: "https://fanout.sh/daily/2026-10-05-self-consistency-reasoning.md"
last_updated: "2026-10-05"
access: "public"
---

# Self-Consistency Improves Chain of Thought Reasoning in Language Models

A model can reach the same answer along several different routes. This paper shows how sampling those routes and voting on their final answers improves reasoning accuracy without additional training, then exposes the extra computation and the gap between agreement and correctness.

## Paper details

- Authors: Xuezhi Wang et al.

- Venue: ICLR 2023

- Track: Inference

- Difficulty: Easy

- Reading time: 28 min

- Original paper: https://arxiv.org/abs/2203.11171v4

## What you will learn

- How self-consistency replaces one greedy chain of thought with several sampled solutions and a vote over final answers.

- Why grouping answers differs from selecting the most probable individual explanation.

- How to read the accuracy gains alongside generation cost, answer extraction, and repeated mistakes.

## Continue exploring

[Read the original chain-of-thought guide](https://fanout.sh/daily/2026-08-18-chain-of-thought): Understand the worked-example prompting baseline before changing how its generated answers are combined.

---
This representation contains public Fanout content only. Protected Pro lessons, account data, billing, checkout, and pricing are not included.

Browse the public content map: https://fanout.sh/sitemap.md