A Node-based proxy intercepts Claude requests and selectively converts dense context such as code, logs, and JSON into PNG images that Fable 5 processes through its vision pathway, exploiting fixed per-image token pricing to shrink input bills on Fable 5 workloads.
I noticed curious price scaling for this myself when working on visual memory last year. The cost to interpret images was not commensurate across methods and platforms.
~60% Fable cost cut by transparently turning the code into an image and having the model OCR it. WILD idea. also hilarious. https://github.com/teamchong/pxpipe
Users report 59–70 percent lower end-to-end costs when the material packs tightly into 1928-by-1928 images, yet the same trick can raise bills on sparse prose where the density edge disappears.
Fable 5 handles gist and arithmetic reliably from the rendered images, but verbatim recall of hex strings or IDs can fail silently, so the repo keeps an escape hatch for critical text that stays uncompressed.
Positive users called the code-to-image hack for slashing Fable AI token costs a brilliant workaround, while negative users dismissed it as exploiting mispriced billing or risking extra OCR errors.
No Digg Deeper questions have been answered for this story yet.
I don't understand this. If it is indeed that much cheaper and not just a mispricing, as the DS paper says it isn't, why aren't all the labs just doing this anyway in the backend to increase margins and cut prices?
~60% Fable cost cut by transparently turning the code into an image and having the model OCR it. WILD idea. also hilarious. https://github.com/teamchong/pxpipe
@EthosVentures https://www.seangoedecke.com/text-tokens-as-image-tokens more here, including a link to the white paper
Developers found a cheaper way to feed Fable 5 large context by showing it pictures of text.
Normally, every code block, log, tool output, and old chat turn becomes text tokens.
Those tokens are billable units.
pxpipe changes the input. It renders dense text into PNG pages, then sends those pages as image blocks.
Fable 5 can read the pixels with OCR-like vision skills, so meaning often survives.
The price gap appears because one image has a mostly fixed token cost.
That cost barely changes when readable text gets packed into the same image.
a 1928×1928 image costs about 4,761 vision tokens.
The same page can hold roughly 92K characters, so dense code becomes cheaper.
The catch is that this is compression through vision, not lossless text storage.
Fable 5 may understand the gist while misreading exact IDs, hashes, names, or strings.
That makes it useful for bulky background context, but risky for byte-exact facts.
~60% Fable cost cut by transparently turning the code into an image and having the model OCR it. WILD idea. also hilarious. https://github.com/teamchong/pxpipe
~60% Fable cost cut by transparently turning the code into an image and having the model OCR it. WILD idea. also hilarious. https://github.com/teamchong/pxpipe
this was in DeepSeek?
~60% Fable cost cut by transparently turning the code into an image and having the model OCR it. WILD idea. also hilarious. https://github.com/teamchong/pxpipe
@dtdev_x maybe @sjgoedecke can tell us? https://www.seangoedecke.com/text-tokens-as-image-tokens was something that came up trying to understand why this works
@EthosVentures read the DeepSeek paper. it's more than a pricing hack, as bizarre as that may be
@MiTypeScript Lol that makes no sense and just means images are mispriced and will soon increase unfortunately
@phillpafford lol, for now. depends if it's cheaper because it's more efficient or cheaper because gaming the pricing model. if the latter.. probably a ToS endgame in sight.
does anyone have a benchmark of how accurate models are at identifying small details and large amts of text in an image?
This is absurdly clever: there is a way you can cut Fable 5 costs by up to ~70%. Just turn Claude Code context into an image and make Fable OCR them. 😂
https://github.com/teamchong/pxpipe
@MiTypeScript @flyosity a picture is worth a thousand tokens
@MiTypeScript lmao someone plz benchmark this
@MiTypeScript This sounds like DeepSeek OCR
the way you know teamchong created pristine one-page exam cheat sheet in school
This is absurdly clever: there is a way you can cut Fable 5 costs by up to ~70%. Just turn Claude Code context into an image and make Fable OCR them. 😂
https://github.com/teamchong/pxpipe
@MiTypeScript How does that even make sense
@MiTypeScript Also see https://blog.can.ac/2026/06/10/snapcompact/ by @_can1357. Works very well in practice!
@MiTypeScript Wow, that is very interesting. Does it actually work?
@krishnanrohit because it changes the semantics from the pov of the model and there's a known eval phenomenon of "text prompts given as vlm images cause overall performance/reasoning degradation compared to naive text prompts"
@MiTypeScript @AnthropicAI @claudeai @elonmusk @sama @OpenAI @xai @DarioAmodei You don't know how to configure it properly, taking someone else's work at this sensitive level, thinking you can simply plug it in as if it was a light bulb, would be foolish of you @DarioAmodei and same for the rest. Matters as this need Delicate care and time.@Google
@MiTypeScript another reason why i think charging for tokens is so stupid from a consumer perspective bc they really should be charging by actual energy consumption like Neuralwatt.
