AI station / mini-PC
RTX 5090 Desktop (32 GB VRAM, 64 GB RAM) — Local LLM Performance
RTX 5090 Desktop (32 GB VRAM, 64 GB RAM) is a Desktop running Windows or Linux, with 32 GB VRAM + 64 GB system RAM.
Specifications
- Memory for models
- 32 GB
- Memory bandwidth
- 1,792 GB/s
- Memory upgradeable
- Yes
- Architecture
- Blackwell GB202
- Form factor
- Desktop
- Operating system
- Windows or Linux
- Released
- 2025
- Llama 3.1 8B Q4_K_M
- 113–234 tok/s
Worth knowing before you buy
- 575 W board power — budget for a 1000 W+ PSU and the heat it puts into the room.
- Models larger than 32 GB must offload to system RAM, which costs roughly an order of magnitude in speed.
Models on the RTX 5090 desktop
- Yi 1.5 FamilyYes, but it is tight
- VibeThinkerYes — comfortably
- Aya 3B (Tiny Aya)Yes — comfortably
- StarCoder 2Yes — comfortably
- SmolLM3Yes — comfortably
- SmolLM2Yes — comfortably
- Qwen3-CoderYes
- Qwen3.8Yes, but it is tight
- Qwen 3.6Yes, but it is tight
- Qwen 3.5Yes, but it is tight
- Qwen 3Yes, but it is tight
- Qwen 2.5 VLYes — comfortably
- Qwen 2.5 FamilyYes, but it is tight
- Poolside Laguna XS 2.1Yes, but it is tight
- Phi-4 MiniYes — comfortably
- Phi-4 FamilyYes — comfortably
- Phi 3.5 FamilyYes — comfortably
- OLMo 2Yes — comfortably
- North Mini CodeYes
- Nex-N2Yes, but it is tight
- Nex-N2.5Yes, but it is tight
- Nemotron Cascade 2Yes
- Nemotron 70BYes
- Nemotron 3 Nano OmniYes
- Mistral Small 3.1Yes
- Mistral Small 3.2Yes
- Mistral FamilyYes
- MinistralYes — comfortably
- Ministral 3Yes — comfortably
- MiniCPM-VYes — comfortably
- Magistral SmallYes
- Llama 3.3Yes
- Llama 3.2 VisionYes — comfortably
- Llama 3.2 FamilyYes — comfortably
- Llama 3.1 FamilyYes — comfortably
- LFM2.5Yes
- InternLM 3Yes — comfortably
- IBM Granite 4.0Yes, but it is tight
- IBM Granite 4.2Yes
- IBM Granite 4.1Yes
- Granite 3.0Yes — comfortably
- GPT-OSSYes — comfortably
- GLM-4.7 / GLM-Z1Yes, but it is tight
- GLM-5 / GLM-5.1No
- Gemma 4Yes
- Gemma 3nYes — comfortably
- Gemma 3Yes, but it is tight
- Gemma 2 FamilyYes — comfortably
- Falcon 3Yes — comfortably
- EXAONE 3.5Yes, but it is tight
- DevstralYes
- DeepSeek R1Yes, but it is tight
- DeepSeek-OCRYes — comfortably
- Cosmos 3Yes — comfortably
- Command R FamilyYes
- Cogito v1Yes, but it is tight
- CodestralYes
- Bonsai 27BYes — comfortably
- BitNet b1.58Yes — comfortably
- Aya ExpanseYes