audio model · whisper · Android
Can I run Whisper large-v3 on Samsung Galaxy S25 Ultra (16GB, 1TB config only)?
Yes. Whisper large-v3 runs on Samsung Galaxy S25 Ultra (16GB, 1TB config only) at int8 (~2.5 GB of ~12 GB usable).
Runs at int8 using ~2.5 GB of ~12 GB usable.
- Peak memory
- ~2.5 GB
- Usable on device
- ~12 GB
- Device memory
- 16 GB
- Quant
- int8
How to run it
Use whisper.cpp or faster-whisper at int8. It is light enough to run on CPU; a GPU just makes it faster.
- Type
- Speech to text
- Parameters
- 1.55B
- Peak memory
- ~2.5 GB at int8
- License
- MIT
- Memory
- 16 GB ram
- Usable for weights
- ~12 GB
- Power draw
- ~8 W
- Best runtime
- llama.cpp (PocketPal) or MLC-LLM (Adreno GPU path)
You could also run
Run Whisper large-v3 on other hardware
FAQ
Can Samsung Galaxy S25 Ultra (16GB, 1TB config only) run Whisper large-v3?
Yes. Whisper large-v3 runs on Samsung Galaxy S25 Ultra (16GB, 1TB config only) at int8 (~2.5 GB of ~12 GB usable).
How much memory does Whisper large-v3 need?
Samsung Galaxy S25 Ultra (16GB, 1TB config only) has room to spare. At int8 the realistic peak is ~2.5 GB of memory.
What do I use to run Whisper large-v3 locally?
Whisper large-v3 runs in whisper.cpp or faster-whisper (among others). It runs on CPU, so no GPU is required.
Sources
- gadgetversus.com
- github.com/a-ghorbani
- github.com/ggml-org
- github.com/google-ai-edge
- github.com/mlc-ai
- github.com/openai
- github.com/shubham0204
- github.com/SYSTRAN
- github.com/Vali-98
- gsmarena.com/samsung_galaxy_s25_ultra-12750.php
- gsmarena.com/samsung_galaxy_s25_ultra-13322.php
- huggingface.co
- layla-network.ai
- sammyfans.com
- sammyguru.com
- samsung.com
VRAM figures are sourced peak-usage anchors at the noted quant, validated 2026-08-03. See methodology.