Local AI model for 8 GB vram

[junaeed365-png]

Can anyone tell me which local AI models are best to run as agent or chatbot assistant on a 8gb vram/16gb ram computer for cybersecurity tasks(abliterated)? Most efficient and accurate with more data.

Environment:
Cachy OS
Ollama-Docker

Hardware Specification:
Core ultra 7 155H
LPDDR5 16 GB RAM- 6400 MT/S
RTX 4070 mobile 8 GB VRAM (125MGP)
1 TB SSD 7300 MB/s
(Acer Predator Helios Neo 14 AI)

Field of study:
Main- Computer Science and Engineering(B.Sc.)
Sub- Machine Learning, Data Science, Cybersecurity(Main focus)

[rayan-sharma-git]

I'd recommend Qwen3 8B Q4_K_M on Ollama. It fits 8GB VRAM well and gives a great balance of coding, reasoning, and chatbot performance for cybersecurity tasks.