Gemma 3 4B IT
4BGoogle Gemma 3
Google Gemma 3 multimodal 4B. 128K context; strong vision + text on 8GB cards.
131K
Max Context
3
Quant Variants
GGUF Q8_0
Best Quality
+0.2%
Perplexity rise (best build)
Quantization Variants
Per-quant VRAM, quality loss, and inference speed on RTX 4090
Measured = run on this site · Estimated = not run here: calculated, or a figure this site has not verified · Community = public reports
| Format | Level | BPW | VRAM | PPL Loss | Speed | Source | Actions |
|---|---|---|---|---|---|---|---|
| GGUF | Q4_K_M | 4.85 | 3.0 GB | 3.2% | 175 tok/s | Estimated | |
| GGUF | Q8_0 | 8.5 | 5.0 GB | 0.2% | 145 tok/s | Estimated | |
| AWQ | INT4 | 4 | 2.5 GB | 4.2% | 210 tok/s | Estimated |
Which file each size is for
A file size is not a memory figure: the VRAM column is the calculator's total at 4K context — weights plus KV cache plus a 10% activation buffer. Units differ on purpose: file sizes are decimal GB, as Hugging Face lists them; the VRAM column is GiB (1024³ bytes) with 2% added to the weights, so 7.12 GB of file counts as about 6.8 GiB there. A vision projector (mmproj), where the repo ships one, is a separate download and is in neither number.
- GGUF Q4_K_MSized at the calculator's generic rate for this level; no file was checked.
- GGUF Q8_0Sized at the calculator's generic rate for this level; no file was checked.
- AWQ INT4Sized at the calculator's generic rate for this level; no file was checked.
Similar models
Compare with Gemma 3Gemma 3 12B IT
Google Gemma 3
Mid-size Gemma 3 with vision. Fits 16GB at Q4; excellent multilingual chat.
Gemma 3 27B IT
Google Gemma 3
Gemma 3 large instruct with long context and multimodal support. At Q4_K_M it needs about 18.3 GB at 4K, comfortable on a 24 GB card. The vision encoder is a separate 0.9 GB file.
Gemma 4 E4B IT
Google Gemma 4
Google's on-device Gemma 4: about 4.7B parameters doing the work plus a 2.8B per-layer embedding table, a 128K window, and image and audio input. Only 4 of its 42 layers keep a cache that grows with context, so the full 128K window costs about 2.0 GB of cache. At Q4_K_M it needs about 5.3 GB at 4K, comfortable on an 8 GB card, and 7.5 GB at 128K, tight on an 8 GB card. An 8 GB Mac gives its GPU about 6 GB, so 5.3 GB at 4K is tight on an 8 GB Mac. Sizes here count the whole model in GPU memory, which is what a Mac or a full GPU load uses; llama.cpp keeps the embedding table in system RAM instead, so on a discrete card real VRAM use is lower than shown. Sizes come from unsloth's files (Q4_K_M 5.0 GB, Q8_0 8.2 GB); other publishers' Q4_K_M files are up to 0.4 GB larger. The image and audio encoder is a separate 1.0 GB file.
Gemma 4 E2B IT
Google Gemma 4
The smallest Gemma 4: about 2.3B parameters doing the work plus a 2.3B per-layer embedding table, a 128K window, and image and audio input. Its cache barely grows — 3 of 35 layers track the full context, about 0.8 GB at 128K. At Q4 it is about 3.0 GB at 4K and 3.8 GB at the full window — comfortable on any card in this index and on an 8 GB Mac. Sizes count the whole model in GPU memory; llama.cpp keeps the embedding table in system RAM, so on a discrete card it needs roughly 1.3 GB less VRAM than shown at Q4. Sizes use the calculator's generic rates — no GGUF file size could be checked.
Running Gemma 3 4B IT locally
At Q4_K_M and 4K of context, Gemma 3 4B IT needs about 3.0 GB — 2.5 GB of weights, 0.25 GB of KV cache and a 0.3 GB activation buffer. The smallest cards in this index that clear that comfortably are 8 GB ones — 11 of them, from the Mac M1 8G up — and 78 of the 78 cards here do. These are calculated figures, not measurements: the estimate stops counting a card as comfortable at 88% of its VRAM, which is roughly the room a desktop session needs.
What longer context costs
Going from 4K to 32K adds about 0.54 GB, taking the total to 3.6 GB. That is gentler than the parameter count suggests: only 5 of 34 layers keep a KV cache that grows with context, and 29 more cap theirs at a sliding window, so the cache scales at a fraction of the usual rate. The model's native window is 128K; holding all of it at Q4_K_M would need about 6 GB.
Which build to download
This index tracks 2 formats for it — GGUF, AWQ — across 3 levels. Q8_0 carries the lowest published perplexity loss at 0.2%. The fastest level listed is AWQ INT4 at about 210 tok/s on an RTX 4090, batch 1 — an estimate, not a run on this site's hardware. GGUF runs on llama.cpp and Ollama across NVIDIA, AMD and Apple silicon; AWQ and GPTQ target vLLM on CUDA and ROCm; EXL2 is ExLlamaV2 — CUDA only, and now archived, so local use rather than serving.
Common questions
- How much VRAM does Gemma 3 4B IT need?
- About 3.0 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 3.6 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.
- Will Gemma 3 4B IT run on a 8GB GPU?
- Yes — at Q4_K_M and 4K context it needs about 3.0 GB, which leaves 3.0 GB spare on a Mac M1 8G, or on any of the other 8 GB cards here (11 in all). That is the smallest size in this index that clears it comfortably; 78 of 78 do.
- Which quantization of Gemma 3 4B IT should I use?
- Q8_0 has the lowest published quality loss (0.2%), and Q4_K_M is the level most people run. All 3 levels in the index are Q4_K_M, Q8_0, AWQ INT4.
Where this model fits
Sized at Q4_K_M with a 4K context window, smallest card first. Comfortable means the estimate uses at most 88% of the memory.
- Runs comfortably on
- Mac M1 8G8GB · +3.0Mac M3 8G8GB · +3.0RTX 40608GB · +5.0RTX 4060 Ti 8G8GB · +5.0Radeon RX 9060 XT 8G8GB · +5.0
- Compared here
- GGUF vs AWQ
- Try it yourself
- Size it in the calculatorGenerate the run command
How to actually run this
Deployment guides for this model and this class of hardware.
This page's figures change when the model or the runtime does.Last updated 2026-11-11 RSS → /feed.xml