Receipt scanner that runs a fine-tuned 0.8B vision model on the phone (free) - detailed cost tracking
I wanted to track grocery spending per item without sending my receipts to anyone, so I built an Android app around a small vision model that reads a photo of a receipt into JSON: store, date, total, every line item. (iOS will take a couple more days to be checked by the Apple team.)
Proof1 community thread
Key points
- Also, most apps only allow broad categories like “Food” or whatever, but I wanted to see a more detailed breakdown, e.g. 150 euros for coffee...
- I fine-tuned Qwen3.5 0.8B and OvisOCR2 (which is itself a Qwen3.5 0.8B finetune) on receipt data.
- (The models used by the app will be published as open source (to use with your own Server) )
- I would appreciate feedback from different devices !
Sources (1)
- [1]Receipt scanner that runs a fine-tuned 0.8B vision model on the phone (free) - detailed cost trackingr/LocalLLaMA (top, daily) · Oct 12, 06:04 AM
I wanted to track grocery spending per item without sending my receipts to anyone, so I built an Android app around a small vision model that reads a photo of a receipt into JSON: store, date, total, every line item. (iOS will take a couple more days to be checked by the Apple team.)
Also, most apps only allow broad categories like “Food” or whatever, but I wanted to see a more detailed breakdown, e.g. 150 euros for coffee...
Extractive summary: sentences quoted from the sources.
Before this
- Oct 10, 2026Satya Nadella says we should assume all AI models are ‘compromised’
- Oct 10, 2026vllm-sr/d3
- Oct 10, 2026[AINews] TypeSafe/Jev at >$100M ARR, $7.5B valuation 3 weeks after launch
- Oct 9, 2026Microsoft's Decision-1 model enters the fast-growing AI decision model race
- Oct 8, 2026[AINews] not much happened today
- Oct 8, 2026Understanding Jev, the new model everyone is talking about
