Gemma 3 on 8GB, 12GB, 16GB and 24GB GPUs

Four graphics cards of increasing size on a workbench

Google's Gemma 3 comes in five sizes: 270M, 1B, 4B, 12B and 27B parameters. The model card lists a 128K-token context for the 4B, 12B and 27B models and 32K for the two smallest, and it describes the family as multimodal, taking text and images as input. On a home GPU the real question is which size fits, and how much room is left once you start feeding it long documents or pictures.

Sizing Gemma 3 with one formula

llmhardware.io sizes a model as parameters in billions × bytes per parameter × 1.2, plus 2 GB for the system. At Q4_K_M a weight takes about half a byte, and at 8-bit it takes one. Run the sizes through it and you get:

  • Gemma 3 1B: about 2.6 GB at Q4.
  • Gemma 3 4B: about 4.4 GB at Q4, 6.8 GB at 8-bit.
  • Gemma 3 12B: about 9.2 GB at Q4, 16.4 GB at 8-bit.
  • Gemma 3 27B: about 18.2 GB at Q4, 34.4 GB at 8-bit.

Card by card

An 8 GB card such as the RTX 4060 runs Gemma 3 4B comfortably, even at 8-bit. It does not hold the 12B at Q4, which needs about 9.2 GB.

A 12 GB card is where Gemma 3 12B at Q4 starts to fit, with roughly 3 GB left for context. Keep the window modest on this tier.

A 16 GB card such as the RTX 5060 Ti 16GB or RTX 4060 Ti 16GB runs the 12B at Q4 with more headroom for long prompts and images. The 8-bit 12B lands at about 16.4 GB, a little over the line, and the 27B misses at Q4 too.

A 24 GB card, like an RTX 3090 or RTX 4090, is the tier for Gemma 3 27B at Q4, with about 6 GB to spare for context. The 32 GB RTX 5090 adds more room for a long window, although the 8-bit 27B, at about 34.4 GB, is still out of reach on a single card.

Images cost context too

Gemma 3's model card says each image is normalized to 896 × 896 pixels and encoded to 256 tokens. One screenshot is cheap. Twenty scanned pages at 256 tokens each add 5,120 tokens before you type a question, and every token in the window grows the KV cache, the memory the model uses to remember what it has already read. If a model that ran fine on text slows down once you attach images, run ollama ps. A PROCESSOR value under 100% GPU means part of the model moved to system RAM.

Which one to pick

For chat and writing help on 8 GB, Gemma 3 4B is the easy choice. On 12 to 16 GB, the 12B at Q4 is the size that fits with room to work. Save the 27B for a 24 GB card or a Mac with enough unified memory, and leave a few gigabytes free for context.

The Ollama tags are gemma3:4b, gemma3:12b and gemma3:27b. In LM Studio, pick the Q4_K_M GGUF of the same size. The Gemma 3 hardware requirements guide has the full table and notes for more cards.

Sources: Google Gemma 3 model card (ai.google.dev/gemma/docs/core/model_card_3), NVIDIA GeForce RTX 30, 40 and 50 series spec pages. Memory figures are formula estimates, not benchmarks.

Comments

Post a Comment

Popular posts from this blog

AMD RX 9070 XT for Local LLMs: ROCm Performance Guide (2026)

The AMD RX 9070 XT launched in early 2026 at $599 with 16GB GDDR6 and 648 GB/s memory bandwidth. For local LLM users who want a mid-range GPU without NVIDIA prices, this changes the calculus significantly. Here is what it actually delivers. RX 9070 XT Specs vs Competition GPU VRAM Bandwidth Price LLM Score RX 9070 XT 16 GB 648 GB/s $599 ~60 tok/s RTX 4070 Ti Super 16 GB 672 GB/s $799 ~62 tok/s RTX 4080 Super 16 GB 736 GB/s $999 ~68 tok/s RTX 4090 24 GB 1,008 GB/s ~$1,100 ~90 tok/s At $599 with 16GB VRAM, the RX 9070 XT is the best value GPU for local LLMs in the mid-range tier in 2026. See the full comparison at LLMHardware.io's AMD GPU guide . ROCm Setup on Linux (2026) # Ubuntu 22.04/24.04 sudo apt install rocm-hip-libraries hip-base # or use the AMD installer: wget https://repo.radeon.com/amdgpu-install/6.3/ubuntu/jammy/amdgpu-install_6.3.60300-1_all.deb sudo dpkg -i amdgpu-install*.deb sudo amdgpu-install --usecase=rocm,lrt,openclsdk # Verify ROCm sees your GPU r...
Read more

llama.cpp flash attention: how to enable and benchmark

Image
Flash attention computes attention with far less memory traffic. The math is identical; the memory movement is not. In llama.cpp it matters most in two situations: long context windows and tight video memory (VRAM). With flash attention enabled, the key-value cache shrinks substantially. A 32k-token context window becomes realistic on a 16GB card, and the prompt-processing phase speeds up, so long documents get ingested faster. The catch is compatibility: not every model and quantization combination supports it, and on some setups the speedup is modest. If you run short chats on a strong GPU, you may barely notice. Where it shines is retrieval-augmented generation (RAG) and long-document work on mid-range cards. Enable it, compare your tokens per second, and keep it if your workload benefits. For the common case of pasting a long PDF and asking questions, it is close to free performance. See how context length changes VRAM needs .
Read more

SillyTavern with Local LLMs: Complete Setup for AI Companions

Image
SillyTavern is the most feature-rich frontend for character-based AI conversations. What began as an AI roleplay tool has evolved into a powerful platform with persona management, memory systems, extensive prompt engineering tools, and connections to every major local inference backend. Here's how to set it up with local models. Quick connect: SillyTavern connects to Ollama, KoboldCpp, llama.cpp server, and TabbyAPI (ExLlamaV2) out of the box. Any of the hardware setups at LLMHardware.io provides a solid backend for SillyTavern. Installation Install via git: git clone https://github.com/SillyTavern/SillyTavern && cd SillyTavern && npm install then node server.js . Access at localhost:8000. For persistent installation, use the included start.sh script. SillyTavern uses Node.js and runs on any platform. The companion app (STMP) enables mobile access to your SillyTavern instance via Tailscale or local network. Character Cards and the TavernAI Format Chara...
Read more