Repository navigation
[BUG] ai is saying that it can't read a pdf. but the model is support it. #9825
Description
Description
I use the opencode deskop app. today downloaded. so it is the latest version.
when using gpt 5.2 the model is refusing to read any pdf. the model supports pdf reading. see
"Only models that support both text and image inputs [...]" https://platform.openai.com/docs/guides/pdf-files
"Input: Text, image" https://platform.openai.com/docs/models/gpt-5.2
In Claude code I can say: "read the pdf" an it reads the pdf.... why is opencode refusing to read pdfs?
Plugins
No response
OpenCode version
latest from 21.01.2026
Steps to reproduce
- open open code app
- use gpt 5.2 thats supports text and image input (aka also pdf)
- try to read in the pdf with the agent ... it refuses ... it rey to extract teh text
Screenshot and/or share link
No response
Operating System
windows 11
Terminal
No response
Metadata
Metadata
Assignees
Labels
No labels
Activity
github-actions commentedon Jan 21, 2026
This issue might be related to existing image/file handling issues. Please check:
These issues suggest there may be underlying problems with how OpenCode handles file/image inputs with GPT-5.2 models.
Feel free to ignore if your specific case with PDF reading is different from image pasting issues.
AndreMonthy commentedon Feb 12, 2026
I have the same issue with Kimi K2.5 via NVIDIA Build API.
kaanaksit commentedon Mar 14, 2026
Indeed this is not a fix.But I strictly state the agent that
include_full_text=trueis needed withpdf-readertool for full text extraction from a PDF. It works when I do that.v1.6.0: fix PDF-reading failure and add anti-capability-lying rule
github-actions commentedon Jul 6, 2026
To stay organized issues are automatically closed after 60 days of no activity. If the issue is still relevant please open a new one.