What can NVIDIA GeForce RTX 3090 run for agents?
Build: NVIDIA GeForce RTX 3090 + — + 32 GB RAM (windows)
Runs comfortably120 models
Ranked by fit for agents use case + predicted speed. Click a row for VRAM breakdown.
Quant: Q8_0Context: 8,192VRAM: 14.7 GBHeadroom: 9.3 GBollama run hermes3:8b72tok/sEstimated
ollama run hermes3:8bQuant: Q5_K_MContext: 8,192VRAM: 19.8 GBHeadroom: 4.2 GBollama run qwen2.5:14b63tok/sEstimated
ollama run qwen2.5:14bQuant: Q4_K_MContext: 2,048VRAM: 19.5 GBHeadroom: 4.5 GBollama run dolphin-mistral:24b42tok/sEstimated
ollama run dolphin-mistral:24bQuant: Q8_0Context: 8,192VRAM: 10.8 GBHeadroom: 13.2 GBollama run qwen2.5:7b82tok/sEstimated
ollama run qwen2.5:7bQuant: Q4_K_MContext: 8,192VRAM: 11.0 GBHeadroom: 13.0 GB126tok/sEstimated
Quant: Q5_K_MContext: 8,192VRAM: 10.7 GBHeadroom: 13.3 GBollama run mistral:7b126tok/sEstimated
ollama run mistral:7bQuant: Q4_K_MContext: 8,192VRAM: 12.2 GBHeadroom: 11.8 GBollama run ornith:9b112tok/sEstimated
ollama run ornith:9bQuant: Q4_K_MContext: 8,192VRAM: 16.3 GBHeadroom: 7.7 GBollama run hf.co/JetBrains/Mellum2-12B-A2.5B-Thinking-GGUF-Q4_K_M83tok/sEstimated
ollama run hf.co/JetBrains/Mellum2-12B-A2.5B-Thinking-GGUF-Q4_K_MQuant: Q4_K_MContext: 2,048VRAM: 19.5 GBHeadroom: 4.5 GBollama run mistral-small:24b42tok/sEstimated
ollama run mistral-small:24bQuant: FP16Context: 8,192VRAM: 19.8 GBHeadroom: 4.2 GBollama run llama3.1:8b38tok/sEstimated
ollama run llama3.1:8bQuant: Q4_K_MContext: 8,192VRAM: 11.1 GBHeadroom: 12.9 GBollama run dolphin3:8b126tok/sEstimated
ollama run dolphin3:8bQuant: Q4_K_MContext: 8,192VRAM: 17.6 GBHeadroom: 6.4 GBollama run qwen3:14b72tok/sEstimated
ollama run qwen3:14bRuns with tradeoffs56 models
Tight VRAM, partial CPU offload, or context-limited.
Quant: Q4_K_MContext: 8,192VRAM: 35.7 GBHeadroom: 7.5 GB- • Partial CPU offload: ~33% of layers run on CPU
ollama run qwen3:30b34tok/sEstimated
- • Partial CPU offload: ~33% of layers run on CPU
ollama run qwen3:30bQuant: Q4_K_MContext: 8,192VRAM: 36.8 GBHeadroom: 6.4 GB- • Partial CPU offload: ~35% of layers run on CPU
ollama run qwen3-coder:30b34tok/sEstimated
- • Partial CPU offload: ~35% of layers run on CPU
ollama run qwen3-coder:30bQuant: Q4_K_MContext: 8,192VRAM: 23.9 GBHeadroom: 0.1 GB- • Tight VRAM fit — only 0.1 GB headroom left for context growth
ollama run qwen2.5-coder:32b31tok/sEstimated
- • Tight VRAM fit — only 0.1 GB headroom left for context growth
ollama run qwen2.5-coder:32bQuant: Q4_K_MContext: 8,192VRAM: 37.8 GBHeadroom: 5.4 GB- • Partial CPU offload: ~36% of layers run on CPU
ollama run qwen3:32b31tok/sEstimated
- • Partial CPU offload: ~36% of layers run on CPU
ollama run qwen3:32bQuant: Q4_K_MContext: 8,192VRAM: 37.8 GBHeadroom: 5.4 GB- • Partial CPU offload: ~36% of layers run on CPU
ollama run qwen2.5:32b31tok/sEstimated
- • Partial CPU offload: ~36% of layers run on CPU
ollama run qwen2.5:32bQuant: Q4_K_MContext: 8,192VRAM: 35.7 GBHeadroom: 7.5 GB- • Partial CPU offload: ~33% of layers run on CPU
ollama run nemotron3:nano34tok/sEstimated
- • Partial CPU offload: ~33% of layers run on CPU
ollama run nemotron3:nanoQuant: Q4_K_MContext: 8,192VRAM: 36.8 GBHeadroom: 6.4 GB- • Partial CPU offload: ~35% of layers run on CPU
ollama run north-mini-code-1.034tok/sEstimated
- • Partial CPU offload: ~35% of layers run on CPU
ollama run north-mini-code-1.0Quant: Q4_K_MContext: 8,192VRAM: 37.3 GBHeadroom: 5.9 GB- • Partial CPU offload: ~36% of layers run on CPU
ollama run glm-4.7-flash33tok/sEstimated
- • Partial CPU offload: ~36% of layers run on CPU
ollama run glm-4.7-flashWhat 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: 64 new comfortable, 73 new tradeoff
- • Qwen 3 0.6B
- • Llama 3.2 3B Instruct
- • Qwen 3 1.7B
- • Gemma 3 270M
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: 103 new comfortable
- • Qwen 3 0.6B
- • Qwen 3 30B-A3B
- • Qwen 2.5 Coder 32B Instruct
- • Qwen3.6 27B
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: 116 new comfortable
- • Qwen 3 0.6B
- • Qwen 3 30B-A3B
- • Qwen 2.5 Coder 32B Instruct
- • Qwen3.6 27B
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 (19 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
—
Even with CPU offload, needs more memory than your VRAM (24 GB) + 60% of system RAM (19 GB) combined.
How to read these numbers
Want a specific benchmark we don't have? Email Contact support and we'll prioritize it.