Gemini Audio and video Ollama

[Yussefgafer]

Problem Description

There is a regression/issue when using the Ollama provider https://ollama.com/v1 with Gemini 3 pro and flash

• When attaching Audio or Video, the model receives the text but completely ignores the attachments (or the attachments fail to upload/encode).
• when i send a big video the sending is very fast because that i think the video is not sending.
• and there is no errors in logcat.
• and the logs in the app is shows the attachment was send.
• The same files work perfectly when using the Google Provider.

using https://ollama.com/v1 as OpenAI compatibility

Can you please fix this

Version Information

• App Version: 1.4.1
• Android Version: 14
• Device: Infinix Hot 50

[54xzh]

Thanks for reporting! There's a partial bug here — unsupported media types (video/audio) are silently dropped without any error when using the OpenAI Chat Completions API, which is misleading.

That said, the root cause isn't something we can easily fix: the Chat Completions API doesn't support video input, and its audio parameter format differs from Gemini's native generateContent API, which Ollama doesn't convert between.

Please use the Google Provider (Gemini generateContent API) to send video or audio to Gemini.

[Yussefgafer]

Ok thank you

but can you add this future

from the https://github.com/Cocolalilal/LastChat

because the internet in Egypt is around 1MB/s That's why I manually delete old images so the context is sent faster.

Could you please add this feature?

[54xzh]

will be added in the next version

[Yussefgafer]

Thank you very much ❤️❤️

[Yussefgafer]

Hello brother can you add feature like a Image OCR but for files

like a old files send to model to compress it lika an OCR 🙃

[54xzh]

Could you clarify what type of files you're referring to?

[Yussefgafer]

any file like l txt file or a pdf i send a big logs to ai and it slow down a sending speed