What can NVIDIA GeForce RTX 3090 run for vision?
Build: RTX 3090 + Ryzen 9 5950X + 64GB DDR4 (used market)
Runs comfortably23 models
Ranked by fit for vision use case + predicted speed. Click a row for VRAM breakdown.
Quant: Q8_0Context: 8,192VRAM: 8.4 GBHeadroom: 15.6 GBollama run gemma4:e4b143tok/sEstimated
ollama run gemma4:e4bQuant: Q8_0Context: 8,192VRAM: 8.4 GBHeadroom: 15.6 GBollama run gemma3:4b143tok/sEstimated
ollama run gemma3:4bQuant: Q8_0Context: 8,192VRAM: 5.1 GBHeadroom: 18.9 GBollama run gemma4:e2b286tok/sEstimated
ollama run gemma4:e2bQuant: Q4_K_MContext: 8,192VRAM: 6.5 GBHeadroom: 17.5 GB240tok/sEstimated
Quant: Q4_K_MContext: 8,192VRAM: 10.0 GBHeadroom: 14.0 GB144tok/sEstimated
Quant: Q4_K_MContext: 8,192VRAM: 10.0 GBHeadroom: 14.0 GB144tok/sEstimated
Quant: Q4_K_MContext: 2,048VRAM: 3.3 GBHeadroom: 20.7 GB530tok/sEstimated
Quant: Q4_K_MContext: 8,192VRAM: 10.1 GBHeadroom: 13.9 GB144tok/sEstimated
Quant: Q4_K_MContext: 8,192VRAM: 11.1 GBHeadroom: 12.9 GB126tok/sEstimated
Quant: Q4_K_MContext: 8,192VRAM: 11.1 GBHeadroom: 12.9 GB126tok/sEstimated
Quant: Q4_K_MContext: 4,096VRAM: 9.3 GBHeadroom: 14.7 GB126tok/sEstimated
Quant: Q4_K_MContext: 8,192VRAM: 9.9 GBHeadroom: 14.1 GB144tok/sEstimated
Runs with tradeoffs12 models
Tight VRAM, partial CPU offload, or context-limited.
Quant: Q8_0Context: 8,192VRAM: 20.4 GBHeadroom: 3.6 GB- • Tight VRAM fit — only 3.6 GB headroom left for context growth
ollama run llama3.2-vision:11b52tok/sEstimated
- • Tight VRAM fit — only 3.6 GB headroom left for context growth
ollama run llama3.2-vision:11bQuant: Q4_K_MContext: 2,048VRAM: 21.9 GBHeadroom: 2.1 GB- • Tight VRAM fit — only 2.1 GB headroom left for context growth
ollama run gemma4:26b-moe39tok/sEstimated
- • Tight VRAM fit — only 2.1 GB headroom left for context growth
ollama run gemma4:26b-moeQuant: Q4_K_MContext: 2,048VRAM: 21.9 GBHeadroom: 2.1 GB- • Tight VRAM fit — only 2.1 GB headroom left for context growth
39tok/sEstimated
- • Tight VRAM fit — only 2.1 GB headroom left for context growth
Quant: Q4_K_MContext: 2,048VRAM: 22.0 GBHeadroom: 2.0 GB- • Tight VRAM fit — only 2.0 GB headroom left for context growth
ollama run gemma3:27b37tok/sEstimated
- • Tight VRAM fit — only 2.0 GB headroom left for context growth
ollama run gemma3:27bQuant: Q4_K_MContext: 2,048VRAM: 22.0 GBHeadroom: 2.0 GB- • Tight VRAM fit — only 2.0 GB headroom left for context growth
37tok/sEstimated
- • Tight VRAM fit — only 2.0 GB headroom left for context growth
Quant: Q4_K_MContext: 8,192VRAM: 20.4 GBHeadroom: 3.6 GB- • Tight VRAM fit — only 3.6 GB headroom left for context growth
ollama run muse-glimmer34tok/sEstimated
- • Tight VRAM fit — only 3.6 GB headroom left for context growth
ollama run muse-glimmerQuant: BF16Context: 8,192VRAM: 27.8 GBHeadroom: 34.6 GB- • Partial CPU offload: ~14% of layers run on CPU
- • CPU is the bottleneck — upgrading RAM bandwidth helps more than VRAM here
5tok/sEstimated
- • Partial CPU offload: ~14% of layers run on CPU
- • CPU is the bottleneck — upgrading RAM bandwidth helps more than VRAM here
Quant: Q4_K_MContext: 8,192VRAM: 47.7 GBHeadroom: 14.7 GB- • Partial CPU offload: ~50% of layers run on CPU
- • CPU is the bottleneck — upgrading RAM bandwidth helps more than VRAM here
ollama run nemotron3:33b2tok/sEstimated
- • Partial CPU offload: ~50% of layers run on CPU
- • CPU is the bottleneck — upgrading RAM bandwidth helps more than VRAM here
ollama run nemotron3:33bWhat if you upgraded?
Hypothetical scenarios. We re-ran the compatibility engine for each.
+32 GB system RAM
~$80–150
Doubles your CPU-offload working set. Helps when models don't quite fit in VRAM.
Unlocks: 161 new comfortable, 79 new tradeoff
- • Qwen 3 0.6B
- • Llama 3.1 8B Instruct
- • Qwen 3 8B
- • GPT-OSS 20B
Upgrade to NVIDIA RTX PRO 4500 Blackwell
see current pricing
32 GB VRAM (vs your 24 GB) plus a bandwidth jump from ~936 GB/s to ~896 GB/s.
Unlocks: 193 new comfortable
- • Qwen 3 0.6B
- • Llama 3.1 8B Instruct
- • Qwen 3 30B-A3B
- • Qwen 2.5 Coder 32B Instruct
Add a second NVIDIA GeForce RTX 3090
~$899
Tensor parallelism splits the model across both cards, effectively doubling VRAM. Bandwidth doesn't double — runs ~1.5× the single-card speed in practice.
Unlocks: 213 new comfortable
- • Qwen 3 0.6B
- • Llama 3.1 8B Instruct
- • Qwen 3 30B-A3B
- • Qwen 2.5 Coder 32B Instruct
Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.
Won't runtop 5 popular models
Need more memory than you have. Shown for orientation.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (38 GB) combined.
How to read these numbers
Want a specific benchmark we don't have? Email Contact support and we'll prioritize it.