Can you confirm that if you use the MCP for markitdown, that it’s not using tokens? My understanding is that it does.
2026-07-01 05:02:32
0
tank0man :
I wrote a web wrapper around it in python so that it functions via a web page with drag N drop functionality or you can browse to the file.. it will out put where you tell it.
2026-07-01 15:19:35
0
Loco Coco :
How do you make this calculation of 70%. Give me a concret example
2026-07-01 17:18:23
0
gdgizmo :
Save millions in tokens 🙄
2026-07-01 05:45:35
1
regisking602 :
Claude needs to increase their tokens instead. They ridiculously low. At least match ChatGPT tokens, given we pay the same amounts.
2026-07-01 05:40:12
1
Papppiiiiii Joe :
I see plenty of videos talking about this. But not a single one shows us how to integrate it 😭
2026-07-01 13:57:38
0
Freddy :
Truthfully, an easier option than installing python etc so you can run this is use Google’s ability to convert files to markdown if you download it from the drive to your hard drive
2026-07-01 07:32:16
0
theopencity :
This is exactly why I built pdfaccessibility .ai/api — PDF to structured Markdown/JSON, with reading order + bboxes for RAG.
2026-07-01 06:27:15
0
hadi.rizvi_093 :
literal lifesaver, you don't even need to mess with the github repo. just use inktomd.com. it converts basically any file to .md in two seconds. i started doing this and my context windows stretch SO much further now, saving like 70% of my tokens.
2026-07-01 12:48:19
1
𝔸𝕝𝕨𝕒𝕗𝕖𝕣 🦇 الوافر :
👍👍👍🌹🌹🌹
2026-07-01 03:25:25
0
To see more videos from user @brandnat, please go to the Tikwm
homepage.