60% Fable cost cut by converting code to images and having the model OCR it

#1

Post by dimitropoulos » Fri, Jul 03, 2026, 3:50 PM UTC

60% Fable cost cut by converting code to images and having the model OCR it
github.com

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#2

Post by dimitropoulos » Fri, Jul 03, 2026, 4:02 PM UTC

there's also a DeepSeek whitepaper on this technique https://www.seangoedecke.com/text-tokens-as-image-tokens

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#3

Post by genxy » Fri, Jul 03, 2026, 4:41 PM UTC

This seems like a pricing hack that burns resources, that when the loophole gets closed the price of OCR will have to rise?

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#4

Post by ricardobeat » Fri, Jul 03, 2026, 4:46 PM UTC

genxy wrote: post #3

This seems like a pricing hack that burns resources, that when the loophole gets closed the price of OCR will have to rise?

It’s not a loophole, it just happens that encoding information as optical tokens is much more efficient than text.

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#5

Post by aabhay » Fri, Jul 03, 2026, 4:53 PM UTC

Ahhh my eyes the vibe coded readme

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#6

Post by samrus » Fri, Jul 03, 2026, 5:17 PM UTC

genxy wrote: post #3

This seems like a pricing hack that burns resources, that when the loophole gets closed the price of OCR will have to rise?

Not really. They arent actually using more resources this way either. This might be a fundamental inefficiency thats being removed

It kinda makes sense too. Because while people do read code word by word, we often "glance over" it and do roughly pattern recognition on it to know what it does. Only homing in on something when we need to answer a specific question. I think humans kinda naturally do this exploit anyway

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#7

Post by lpellis » Fri, Jul 03, 2026, 5:18 PM UTC

I tried the same thing last year (with openai models), back then it worked to reduce prompt tokens, but you needed way more completion tokens, ultimately more expensive (and slower) https://pagewatch.ai/blog/post/llm-text-as-image-tokens/

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#8

Post by guardiangod » Fri, Jul 03, 2026, 5:26 PM UTC

ricardobeat wrote: post #4

genxy wrote: post #3

This seems like a pricing hack that burns resources, that when the loophole gets closed the price of OCR will have to rise?

It’s not a loophole, it just happens that encoding information as optical tokens is much more efficient than text.

Truly a picture is worth a thousand words.

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#9

Post by mpalmer » Fri, Jul 03, 2026, 5:31 PM UTC

aabhay wrote: post #5

Ahhh my eyes the vibe coded readme

What, you don't like your caveats to be honest?

Re: 60% Fable cost cut by converting code to images and having the model OCR it

#10

Post by aabhay » Fri, Jul 03, 2026, 5:32 PM UTC

In Gemini at least, if you look at how they process PDFs, they do an OCR and then feed the text + image to the model, without charging you for the text tokens (I believe).

So my guess is that Claude’s backend is doing the same — so this hack is probably more of a loophole in token accounting that might get closed if Claude is doing what Gemini does