canirunaimodel
Google · 33B · Dense
Gemma 4 flagship base model (official)
0 downloads
0 likes
2026-04 256K context
Use Cases
vision reasoning
| Quant | Bits | VRAM | Quality | Status |
|---|---|---|---|---|
| Q2_K | 2 | 11.1 GB | low | — |
| Q3_K_M | 3 | 15.3 GB | moderate | — |
| Q4_K_M | 4 | 17.4 GB | good | — |
| Q5_K_M | 5 | 21.6 GB | good | — |
| Q6_K | 6 | 25.9 GB | excellent | — |
| Q8_0 | 8 | 34.3 GB | excellent | — |
| F16 | 16 | 68.1 GB | lossless | — |
About this model
HF model card
Gemma
33B
Dense model for vision, reasoning workloads.
Context
256K
Q4 VRAM
~17.4 GB
Dense Vision
Gemma 4 31B on your hardware
Gemma 4 flagship base model (official). This page turns the Hugging Face model card into practical local-run numbers, so you can compare quantized VRAM, system RAM, and expected fit before downloading a large checkpoint.
Model shape
33B total parameters. Dense models use the whole network for each token.
Local fit target
Start with Q4_K_M: about 16.1 GB on disk and 17.4 GB VRAM before extra context and runtime overhead.
Best use cases
vision, reasoning
License and source
Listed as Gemma from huggingface.co/google/gemma-4-31B.
Want the real verdict? Pick your GPU or edit the specs on this page and compare the quant table below.
Open HF repo