Ugh, hashtag optimizing is soul-sucking, but you do not need to hand your photos to some random web service to get tags. Run a tiny image-caption/tagging model locally and convert the results into hashtags. My go-to is clip-interrogator, it runs on your own machine, spits out concise keywords and prompt-like descriptions, and is literally made to extract useful phrases from images. Github: github.com/pharmapsychotic/clip-interrogator

If you want something more generic, use BLIP (Salesforce) image captioning via Hugging Face, e.g. the Salesforce/blip-image-captioning models. Install with pip, run the model to get a caption, then use a simple POS filter or spaCy to pull nouns/adjectives and prefix them with #. That workflow is trivial to script and keeps everything local. Models will run on CPU but are far faster with a GPU.

If you absolutely need Android, use Termux and a lightweight model or run the model on a tiny home server and call it from your phone, do not upload to third-party servers. For anime art, DeepDanbooru is the standard local tagger. For everything else, clip-interrogator + a tiny post-processing script is your best bet.

Honestly, stop treating hashtags like SEO black magic, generate sensible descriptive tags locally, and move on with your life. If you want, I can paste a minimal Python snippet that takes an image, runs BLIP or clip-interrogator, and outputs a ready-to-paste list of hashtags. Which model do you want to try, BLIP or clip-interrogator?