---
title: "Discussion on \"n8n Learning: LLM Chain vs Single Call — Caption Generation Experiment\" | Hashnode"
description: "Discussion on \"n8n Learning: LLM Chain vs Single Call — Caption Generation Experiment\". I compared Claude Haiku 3-step chain vs Sonnet single call for Instagram caption generation in n8n. Learn why context continuity beats cost optimization for"
url: "https://hashnode.com/posts/n8n-learning-llm-chain-vs-single-call-caption-generation-experiment/698762d07e645ee90a45b272"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [n8n Learning: LLM Chain vs Single Call — Caption Generation Experiment](https://ai-workflow.hashnode.dev/n8n-learning-llm-chain-vs-single-call-caption-generation-experiment?utm_source=hashnode&utm_medium=feed)

I compared Claude Haiku 3-step chain vs Sonnet single call for Instagram caption generation in n8n. Learn why context continuity beats cost optimization for

Feb 7 · 5 min read · ai-workflow.hashnode.dev  
Updated Feb 16  
Read the article

-   [#n8n](https://hashnode.com/tag/n8n)
-   [#ai](https://hashnode.com/tag/ai)
-   [#automation](https://hashnode.com/tag/automation)
-   [#llm](https://hashnode.com/tag/llm)

## Written by

[masaya ohsugi](https://hashnode.com/@masaya135)

@masaya135 · Nara, Japan

  
[View profile](https://hashnode.com/@masaya135)

From Scissors to Syntax. 20-year Hairdresser building AI SaaS agents.

AI Engineer based in Nara, pivoting from a 20-year career as a hairdresser. Currently building Stylus (AI captioning for salons) and Harmony (Instagram automation). Obsessed with workflow automation using n8n, Python, and LLMs. My mission is to solve inefficiencies in the beauty industry through code. Writing about SaaS development, automation, and my journey from the salon floor to software architecture.

## Responses

Markdown supported · ⌘↵ to post

No responses yet.