firewall disabled on windows.
android and windows are in same wifi network.
windows 11
android 13

cmd command:
ollama run serve

Host in App:
http://192.168.1.103:11434

error:
Issue: Invalid Host

I don't know what to type in the host and I need some help

Thanks for making this great app!

One thing I noticed, it seems it is requesting the ollama server to immediately clear the model from VRAM again after each request. This means there is a lengthy waiting period for every followup question while the model is loaded back into memory. Could that be configurable? I would set the keep_alive in the api request to -1 so it only gets removed from VRAM when it needs to make space for a different model.

PS I noticed this especially when testing the llava visual model. Which does work great by the way!

as title says...

First, thanks a ton for developing this app. It has allowed me to completely move off of OpenAI now.

One request I would make is can additional model details be shown in the UI?

For example I have three flavors of Llama 3 downloaded, but it is hard to determine which is which within the app.

Thanks for your consideration!

I have Ollama running on my Windows machine and ngrok running with a static url. I use that url in the settings of the app and it says it's good, I check the url on multiple devices and it says "Ollama is running!", but when I select the model in the app and send a message I get "Issue: Request Failed. Server issues". Is this a problem with my pc setup or my phone?

I think you made a typo in the system message, do you mean to write "like" instead of "lite"?

Requirements

• I have searched the issues of this repository and believe that this is not a duplicate
• I can confirm this feature request is not already implemented in the app

Description

Could you please consider releasing it to the Play Store?

Possible solution

Self-descriptive

Additional context

My phone has MDM configured, and I can't sideload APKs

Suggestion: Adding openai API compatibility would be much appreciated.

E.g
localhost:port/v1

/models

Etc

open it up to connecting to the powerhouses: llamacpp, localllm and the rest of the local llm openai api supporting software

Cheers

Requirements

• I have searched the issues of this repository and believe that this is not a duplicate
• I can confirm this feature request is not already implemented in the app

Description

As prompt responses can be quite fast, I'd suggest letting the user scroll while the prompt is being generated. This is for instance possible with the official ChatGPT app. It's a quality of life improvement since, if a very large rresponse gets generated, one can not read until the LLM has finished with it's response.

Possible solution

Not sure how it works but possibly disable auto scrolling with new content?

Additional context

No response

I am completely new to this as I just read about it on reddit. Can someone help me how to connect it to the host or what are the steps to follow for it to work properly as i don't know what to do.

When I add my host url it just spins.

add support stop button when generating text

Seen on: https://github.com/JHubi1/ollama-app/tree/1.0.1

You can type messages to any model, and responses are provided for up to a certain number of messages (in my case 6/7).
After that, the model responses shown in the chat are empty.

Other clients and the CLI are not affected.

I understand it is on izzyondroid store. Can this app be added to the f-droid store?

I just wonder why Ollama requests the permission to access the microphone – as neither description nor screenshots give a clue:

! repo/com.freakurl.apps.ollama_8.apk declares sensitive permission(s):
  android.permission.RECORD_AUDIO

Could you please clarify? Thanks in advance!

when AI like Phi3 generating a code block it has cannot word wrap
therefore some of code text is hidden.

It'd be great, if app was consistent with the majority of the android apps out there)

Do I need a computer and connect to it because I don't have one or what do I put in the local host thing btw I'm kinda new on GitHub soo please go easy on me I need a simple explanation if possible thanks

Hey, i installed the app yesterday, added my ollama host, which is running on my local server, so i entered ip:port.
Then the app proceeded, i was able to see the models i have available and i can select them and started to chat, but when i chat, nothing happens. I see a loading circle but never anything comes up.
In the ollama container i can see the api request, but nothing more, gpu doesnt start and no reply comes back.

I recently began messing around with model files to make dedicated custom AI chatbots using Ollama. It works great and was entertaining. I then tried to use that same model with the android client, and found that it doesn't work at all.

I imagine you are overwriting the SYSTEM flag via the API call. I'll probably fork and look into it, but the reason I made a Issue is because I then noticed that the Chat Descriptions for previous chats DO act like the character AI... This is a little unexpected for how I understood the API to work.

For demonstration, here is the relevant chat log from my conversations with a custom Snoop Dogg. AI.

Conversation did not act like expected

But conversation titles are definitely influenced

Sorry another feature request :)

It would be great to be able to turn off the haptics in streaming mode, because my server is pretty fast that so turns into an annoying rumble when I use it. I think it might be more useful with really slow servers.

like png txt or other files
ollama supporting it

L'application est super, mais après avoir envoyé 3 chats, il y a un message d'erreur qui dit "server issue".

I would like to contribute to the translation of the app into Italian, I created a file app_it.arb but I don't know how to insert it

Hi, may I know how much RAM is needed, I'm looking at the 24gb ram phone of OnePlus , I'm just wondering if it's gonna be speedy there or not.

First of all, Thank you for this app. It works great and the UI is top notch.
I am using your app for the last 4-5 hours particularly for chat and It works great.
I tried vision model(llava-phi3) now, It doesn't seem to work for me. I tried it 4-5 times again and again. Restarted the app, restarted my PC and nothing seems to work for vision models.

Here's a screenshot:

It's stuck there forever. But, llava-phi3 is working on my PC perfectly.
Any Ideas?

nice straight forward app :) you could extend compatibility with llama.cpp pretty easily, ollama just wraps llama.cpp, a lot of people use llama.cpp rather than ollama

Currently having a couple of problems, I mostly use 8192 context, and often very long prompts, these fail to send (Nothing like 8192 tokens, but 500-2k) and break the chat

Chats break after some time, not sure why, keep awake set, but even manually nudging the server doesn't make a difference - I don't have this issue with llama.cpp or ollama wrapper

copy/paste is over complicated
chat content lost when a chat breaks

I realise this isn't the purpose of this app, its more chat, but it could be more widely useful, couple of suggestions that could massively extend usefulness (eg for programming tasks) beyond chats

Single tap a message to Copy to Clipboard
Single tap to paste clip board into the input text box

Save full chat (seems like all chat content is lost, if chat breaks?)
Export full chat (a string to the clip board works too)

https://www.virustotal.com/gui/file/359659413b18b45270b5c76665b529f9e9eb0519dcafe074e70d5d6466e7f33f/detection

Hi! I really love the new voice option in 1.1.1! Thanks for writing it.

The speech recognition works very well, it recognises my prompts quickly. But the responses are not played back. Perhaps it's because you expect google TTS? I have the Samsung one selected on my phone.

Also, I noticed that in voice mode when the response is larger than the screen, I can't scroll down to see the rest. But perhaps that would happen automatically when the speech would get there?

Might be just my issue, and im missing smth obvious but i simply cannot connect to the host.
Im using the ollama docker and have forwarded the port 0.0.0.0:11434 -> 11434/tcp so it listens on the host. But whatever I do i cannot get the host to connect.
Any pointers on what i am doing wrong? Would love to give this app a try vs the openwebui.

here is the tutorial:
https://xdaforums.com/t/run-ai-mistral-7b-llm-on-xiaomi-13-ultra-ai-local-install-working-offline.4658021/

as title says...

it's not a issue , but I don't know to how to add the port. Like port do I have to add it would be great if there is a video explanation for this, because I found ollama a opensource to be in our mobile but I kinda feel myself so dumb that I don't know to how to add this port 😅

Hello,

Found this app on reddit. Looks nice and native. However, I can't connect to my server. It just says invalid host on the app. If I connect using a web browser in my LAN, however, I CAN connect so I imagine I am typing my url in right but doing something else wrong?

Any help with this is appreciated!

Have faced this issue with some other FOSS Flutter apps as well that they don't respect the refresh rate set by the device (in my case, it is forced to 120Hz).

Example issue and fix:
emavgl/oinkoin#16 (comment)

Device: Nothing Phone (2)
Android: 14

I think that using ollama-app as a digital assistant instead of google assistant on android would be nice. Android allows 3rd party apps to register as digital assistant so they can be invoked by holding home button or using a screen gesture. It would be perfect if users will be able to hold the home button on any screen and use voice recognition to ask LLM. Or maybe even ask LLM about screen content.

The settings page checks if the response to https://example.com is "Ollama is running". i have a setup where example.com is hosting an ollama web frontend and the reverseproxy proxypasses example.com/api to the ollama service. the result is, that the ollama app refuses the host, but it could use it

I am going to be building a github action that builds and publishes the ollama app automatically.

I will be maintaining a fork of this repository where you can check the progress: https://github.com/turbomaster95/ollama-app

I've set up my Ollama server behind a secure Nginx proxy that checks the Authorization header and does API key verification. I've seen other tutorials online about securing the Ollama API via basic authentication as well. It would be useful to be able to use add an advanced server configuration section to set arbitrary headers when connecting to the server

There is no way to differentiate between for example llama3 and llama3:70b.
They are both being displayed as "llama3" in the app!

Very cool app btw., appreciate your work!