Verified reference stack

Consultant — Confidential Documents — RTX 4070

Official Google material reports Gemma 3 12B int4 weights around 6.6 GB. Hardware score still assumes normal single-user inference and bounded context.

Application

AnythingLLM

Runtime

Ollama

Model

Gemma 3 12B IT (QAT/int4 profile)

Hardware

RTX 4070 12 GB / 32 GB RAM

GOOD

Privacy

Fully local configuration

License / terms

Gemma Terms of Use — professional/legal-entity use allowed subject to terms and prohibited-use policy

Evidence

https://anythingllm.com/download
https://ai.google.dev/gemma/terms
https://developers.googleblog.com/gemma-3-quantized-aware-trained-state-of-the-art-ai-to-consumer-gpus/

Freshness rule: Material changes set this stack to REVALIDATION_REQUIRED.