text model · Mistral · Android

Can I run Ministral 3 3B on Samsung Galaxy S26 Ultra (16GB, 1TB config)?

Compatibility verdict VRAM threshold engine

Yes, it runs usable speed ~21 tok/s est.

Yes. Ministral 3 3B runs on Samsung Galaxy S26 Ultra (16GB, 1TB config) at Q4_K_M (~3.2 GB of ~12 GB usable).

Needs ~3.2 GB Device usable ~12 GB

Runs at Q4_K_M using ~3.2 GB of ~12 GB usable. You have room for FP16 for higher quality.

That figure is at a 4k context and moves about ±15% as context length changes. Samsung Galaxy S26 Ultra (16GB, 1TB config) leaves ~8.8 GB of headroom, room to step up to FP16 for higher quality.

Q4_K_M needed: ~3.2 GB
Usable on device: ~12 GB
Device memory: 16 GB
Best quant: Q4_K_M

Share on X Share on Reddit

Which quant fits

Quant ladder vs ~12 GB usable

Q2_K

~2.5 GB

Q3_K_M

~2.7 GB

Q4_K_M

~3.2 GB

Q5_K_M

~3.3 GB

Q6_K

~3.7 GB

Q8_0

~4.6 GB

FP16

~8.9 GB

The line marks Samsung Galaxy S26 Ultra (16GB, 1TB config)'s ~12 GB budget; rungs past it are too large.

How to run it

On Android use PocketPal AI (Polished app, download GGUF and run offline.).

Model Mistral

Parameters: 3B
Q4_K_M size: 2 GB
Q8_0 size: 3.4 GB
Context: 256k
Ollama tag: ministral-3:3b

Full Ministral 3 3B requirements →

Device Android

Memory: 16 GB ram
Usable for weights: ~12 GB
Best runtime: llama.cpp (PocketPal) or MLC-LLM (Adreno GPU path)

Best models for Samsung Galaxy S26 Ultra (16GB, 1TB config) →

You could also run

Run Ministral 3 3B on other hardware

Nvidia GeForce RTX 2060 (6GB)Apple M1 (8GB)8GB RAM Laptop (CPU/iGPU only)iPhone 15 Pro iPhone 16 iPhone 16 Pro

FAQ

Can Samsung Galaxy S26 Ultra (16GB, 1TB config) run Ministral 3 3B?

Yes. Ministral 3 3B runs on Samsung Galaxy S26 Ultra (16GB, 1TB config) at Q4_K_M (~3.2 GB of ~12 GB usable).

How much memory does Ministral 3 3B need?

Samsung Galaxy S26 Ultra (16GB, 1TB config) has room to spare. At Q4_K_M the weights are ~2 GB; with KV cache and runtime overhead, budget ~3.2 GB at a 4k context.

What is the best tool to run Ministral 3 3B on Android?

On Android, PocketPal AI (Polished app, download GGUF and run offline.) is the go-to option. NPU acceleration is limited and chip-specific; most apps run on CPU. Expect 1B-4B class.

Embed this

A live badge for your model card or README, updated as the data is.

Markdown

[![Ministral 3 3B on Samsung Galaxy S26 Ultra (16GB, 1TB config)](https://localmodel.run/badge/ministral-3-3b/samsung-s26-ultra.svg)](https://localmodel.run/can-i-run/ministral-3-3b/samsung-s26-ultra)

Sources

Weights are measured from GGUF files; KV cache and overhead are computed, so totals can vary ~15% with context and runtime. Any tok/s is a bandwidth estimate. See methodology.