# PSA for Pro users trying Fable this week: it uses your limit about 4x faster than Opus, not 2x — retained record

> Retained reddit record from reddit:ClaudeCode.

Canonical: https://highsignal.app/data/records/8b2bfe4952f96c34

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeCode)  published 2026-11-10 22:57 ingested 2026-11-11 02:42

[original source ↗](https://www.reddit.com/r/ClaudeCode/comments/1xt1eo1/psa_for_pro_users_trying_fable_this_week_it_uses/)

## retained text



```
I think it's natural to assume that as per API pricing, Fable 5.2 should be 2x more expensive than Opus 5.5 ($8/$40 vs $4/$20 per million), but in actual usage limits, it's not. It's 4x more.

So I've been logging my Claude Code usage properly with my own tool I made called 'Claudemeter'. I've got historical values for my entire Pro usage for the last \~8 months or so, and I've seen the actual fluctuations in the usage limits across weeks (might do a separate breakdown on that later if anyone's interested).

Anyway, now that Fable 5.2 landed, I did some tests, and was horrified at how quickly it chewed through the 5-hour limit and 25% of the 'Fable' limit in a simple task, so I had to dig in and do the math.

On 2 similarly long Claude Code sessions, this is what my data looked like:

|Criteria|Opus 5.5|Fable 5.2|
|:-|:-|:-|
|5-hour limit used|91%|68%|
|API-equivalent cost|$33|$13|
|Cost per 1% of the 5-hour limit|$0.36|$0.19|
|Tokens per 1%|1.07M|0.26M|

In other words, after extrapolating that, a full 5-hour window is worth roughly $20 of Fable vs $40 of Opus (and my Opus average over the past month is closer to $44 per 5-hour window).

And since Fable costs 2x per token, the per-token burn rate is roughly 4x more than Opus.

**Bottom-line:** Fable drains usage limits 4x faster than Opus on subscription plans. I can confirm this for Pro, and it seems similar for Max/Ultra plans too based on others' posts (like Theo's post).

Cost-efficiency per task and what value you get out of Fable are separate tests of course and I didn't run those since they already exist. But personally, I could get more work done using Opus 5.5 than with Fable at this rate.

I guess the simple Q to ask is: why can't Anthropic give us the same dollar-value for Fable? Or at least tell us the actual ratio, instead of just "uses limits faster"?
```



document fetched 2026-11-11 00:40 · canonical https://reddit.com/r/ClaudeCode/comments/1xt1eo1/psa_for_pro_users_trying_fable_this_week_it_uses
