Can I Run Gemma 3 on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)?
作者： Jakub Rusinowski · 最后更新： 2025年3月12日
Yes
Yes — Gemma 3 27B Instruct at Q5_K_M needs about 21.3 GB of the 24 GB usable on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM) (~2.7 GB spare), at ~34.6 tok/s (estimated), with room for about 32,768 tokens of context.
Confidence: high · Recommended quantization: Q5_K_M · Estimated speed: ~34.6 tok/s
See what else this hardware can run →
或在 Vast.ai 比较，低至 $0.35/小时 (typical low · varies)
作为亚马逊联盟成员，我们从符合条件的购买中获得收入。云 GPU 链接为推荐链接——我们可能获得佣金，您无需额外付费。
RTX 4090 Desktop (24 GB VRAM, 64 GB RAM) — what it gives a model
| Usable memory for models | 24 GB |
| Memory bandwidth | 1008 GB/s |
| Form factor | Desktop |
| Operating system | Windows or Linux |
| Memory upgradeable | Yes |
Gemma 3 on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM): memory by quantization
| Quant | Memory needed | Fits 24 GB? | Max context | Est. speed | Download |
|---|---|---|---|---|---|
| F16 | 56.1 GB | ✗ No | — | — | 54 GB |
| Q8_0 | 30.8 GB | ✗ No | — | — | 28.7 GB |
| Q6_K | 24.3 GB | ✗ No | — | — | 22.1 GB |
| Q5_K_M | 21.3 GB | ✓ Yes | 32K | ~34.6 tok/s | 19.1 GB |
| Q4_K_M | 18.4 GB | ✓ Yes | 64K | ~39.8 tok/s | 16.3 GB |
| Q3_K_M | 13.6 GB | ✓ Yes | 128K | ~53.3 tok/s | 11.5 GB |
| Q2_K | 11 GB | ✓ Yes | 128K | ~65.4 tok/s | 8.9 GB |
Which Gemma 3 sizes fit
| Variant | Needs at Q4_K_M | Fits? | Est. speed |
|---|---|---|---|
| Gemma 3 27B Instruct | 18.4 GB | ✓ Fits | ~39.8 tok/s |
| Gemma 3 12B Instruct | 9.1 GB | ✓ Fits | ~77.8 tok/s |
| Gemma 3 4B Instruct | 3.6 GB | ✓ Fits | ~172.6 tok/s |
| Gemma 3 1B Instruct | 1.5 GB | ✓ Fits | ~322.9 tok/s |
RTX 4090 desktop limitations
- 24 GB is the sweet spot for 27–32B models at Q4; 70B needs offload or a second card.
Recommended setup
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
How these numbers are calculated
- Memory need = quantized weights + KV cache at 8,192 tokens (f16) + 0.8 GB runtime overhead.
- 24 GB of VRAM on the NVIDIA GeForce RTX 4090 at 1008 GB/s.
- 64 GB of system RAM available for CPU offload when a model exceeds VRAM.
- Throughput is a memory-bandwidth roofline estimate, not a measurement. It is labelled "estimated" everywhere it appears.
- KV cache is computed from this model's published attention configuration.
FAQ
Can I run Gemma 3 on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)?
Yes — Gemma 3 27B Instruct at Q5_K_M needs about 21.3 GB of the 24 GB usable on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM) (~2.7 GB spare), at ~34.6 tok/s (estimated), with room for about 32,768 tokens of context.
Which quantization of Gemma 3 should I use on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)?
Q5_K_M — it needs about 21.3 GB of the 24 GB available, downloads as roughly 19.1 GB, and runs at an estimated 34.6 tokens/sec with up to 32K of context.
What limits Gemma 3 on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)?
Nothing binding — the model fits with headroom and generates at a usable speed on this hardware.
Which runtime should I use?
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
Other Computers
Other Models on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)
- Gemma 3n on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)
- Gemma 4 on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)
- GLM-4.7 / GLM-Z1 on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)
- GLM-5 / GLM-5.1 on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)
- GPT-OSS on RTX 4090 Desktop (24 GB VRAM, 64 GB RAM)
Gemma 3 on GPUs
- Gemma 3 on NVIDIA GeForce RTX 5090
- Gemma 3 on NVIDIA GeForce RTX 5080
- Gemma 3 on NVIDIA GeForce RTX 5070 Ti
- Gemma 3 on NVIDIA GeForce RTX 5070