---
title: "Drew Breunig: Fable's High Cost Marks the End of the Free Lunch in AI Coding"
source: "Simon Willison 博客（Atom）"
source_tier: "t2"
url: "https://simonwillison.net/2026/Aug/23/drew-breunig/"
permalink: "https://feed.astralor.com/items/itm_02dbe79386a6916414afc264"
published: "2026-08-23T19:55:30.000Z"
score: 73
channel: "ai"
tags: ["AI Coding","Anthropic","Claude","Fable","LLM Pricing"]
---

# Drew Breunig: Fable's High Cost Marks the End of the Free Lunch in AI Coding

## AI commentary

> This quote is worth reading for anyone relying on large models for coding. It reminds us that when the new generation of models is no longer 'cheap and powerful', engineering thinking must return: categorize tasks and let the right model do the right work. For AI infrastructure teams, this is a signal that cost optimization and model routing will become as important as model capability.

## At a glance

Before Fable, it felt silly to waste time improving your coding harness or context strategies, because a new model would arrive at the same price or cheaper and paper over most problems. But Fable changed that calculus. While Fable is incredible, its cost is so high that Opus, 5.6, K3, and even GLM are good enough for most code needs. Drew Breunig notes that teams are now thinking about which work should go where, marking the end of the free lunch in AI coding.

## Key points

- New models no longer automatically deliver broad improvements at lower prices; Fable's high cost forces developers to reassess tooling investments.
- Models like Opus, 5.6, K3, and GLM are good enough for most coding tasks, making Fable worthwhile only in specific scenarios.
- Developers are now routing workloads across models instead of blindly chasing the newest and strongest model.

## Full text

Prior to Fable, it felt silly to waste too much time improving your coding harness or context strategies. A new model would arrive at the same price (or cheaper!) and paper over most of your problems. But then Fable landed. It was (and still is!) incredible. But the cost was so high and Opus was good enough (as was 5.6, K3, and even GLM) for most of the code we needed. So we started to think about what work went where. — Drew Breunig, Fable & The End of the Free Lunch Tags: drew-breunig, anthropic, claude, llm-pricing, ai, llms, generative-ai, claude-mythos-fable

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_02dbe79386a6916414afc264