# Fable GUZZLES tokens — retained record

> Retained reddit record from reddit:ClaudeCode.

Canonical: https://highsignal.app/data/records/e2f90c4bbb2b7f37

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeCode)  published 2026-11-11 05:27 ingested 2026-11-12 02:42

[original source ↗](https://www.reddit.com/r/ClaudeCode/comments/1xt9i6x/fable_guzzles_tokens/)

## retained text



```
The model seems excellent so far, but if you let it, it will burn a ton of tokens thinking about a problem from every possible angle (I'm not complaining mind you). Based on the benchmarks from Anthropic as well as outside sources like AA and Vals and others, I thought that moving from Opus 5.5 high to Fable 5.2 medium would provide a little bit of an intelligence boost for a similar cost, but so far Fable seems to be consuming 50-100% more tokens than Opus did. I am on the Ultra plan and push it pretty hard, I usually approach the weekly limit with Opus 5.5 but often don't hit it. With Fable I think I could hit the limit in three days, maybe less. Anyone tried 5.2 low as their base model? If that doesn't work, I'll sadly be heading back to Opus 5.5 (but only after burning through the Fable resets). 
```



document fetched 2026-11-12 00:43 · canonical https://reddit.com/r/ClaudeCode/comments/1xt9i6x/fable_guzzles_tokens
