Hi, I’ve tried a few apps without success to get Gemini, my primary phone LLM assistant, to work on my pebble time 2 with no success. Has anyone found a working approach?
The first app I tried “Gemini voice and text” requires setting up an API and apparently a TTS key. I think I got the first working but second seems to require a paid subscription to Gemini which is where I stopped.
Second that I tried PebbleAI was created in September 2025 and would not connect to the API key I created, it seems to be calling Gemini 2.0 model.
“Gemini voice and text” works well once set up.
Yes, it costs money, but I think this is the only realistic way it’s going to work on your watch, as API calls need to be made. Until someone comes up with a free alternative, this works well.
You get Pixel Watch levels of cool factor when using it.
I just put the minimum amount into the Google Cloud service AU$20.
Based on my brief use, the 15 requests I made cost 7 cents in charges for general knowledge-type queries.
That’s with a mix of Gemini Flash 2.5 and 3.5, as I tried it out.
That equates to about 285 queries before it’ll be used up.
Sticking to the cheaper models will stretch that beyond 300, but it should give you a baseline idea of how much use that’ll equate to.
I realised I made a calculation mistake in my last post; I forgot to multiply by the number of requests I’d made. So my AU$20 will actually get closer to 4500 voice responses to my queries before depleting.
As to Booster’s point, after some searching, it appears the free tier would still be enough if all you want is text responses. Gemini Flash 2.5 and 3.5 allow 250 queries per day for free, more than enough for watch use.
I’ll have a look into forking this app and making just a text-based version that’d work for free.
I’m no Pebble dev, but I’ve been dabbling. A relatively simple alteration should be possible.
If I get anywhere with it, I’ll let you know!
Did a bit more research, seems the voice dictation part requires a speech to text service, before even sending the query to Gemini.
Currently that’s seemingly done via rePebble STT service which costs the app dev $3/month to have.
I’ve shot the “Gemini voice and text” dev a message to see if he can implement a toggle to disable the charged voice part.
As the app already works with Pebble’s that have no mic (and thus is free to use), a simple toggle switch in the app to disable the voice reply on Pebbles that do have a mic, making it essentially free for the end user, would be the best outcome.
I’ll let you know if I get a reply.
I got word back from the developer. He’s pushed out version 1.4.3, which adds a toggle in the app settings to disable the text-to-speech (TTS) feature.
Effectively allowing you to use it for free with just text responses!
