Graphics cards
Which AI models run on the 2x RTX 5090 (64 GB)?
The 2x RTX 5090 (64 GB) has 64 GB of memory, about 62 GB of it usable for a model, and 1792 GB/s of memory bandwidth. 93 of the 110 models in our catalog run on it at 4-bit or better.
Data as of 2 November 2026
64 GBmemory
~62 GB usable for a model
1792 GB/sbandwidth, sets the speed
93models run at 4-bit or better
Two cards in one PC, model split across both. Each token still reads at one card's bandwidth.
Our pick for this machine Llama 3.3 70B The largest model on our shortlist of current models that runs at 4-bit or better and 10 tok/s or more.Models that run on it
| Model | Best build | Memory | Speed |
|---|---|---|---|
| Llama 3.3 70B70.6B | Q6_K | 60 GB | ~17 tok/s our est. |
| Llama 3.3 Nemotron Super 49B49B | Q8_0 | 54 GB | ~19 tok/s our est. |
| Qwen3.6-35B-A3B35B MoE | Q8_0 | 41 GB | ~234 tok/s ModelFit est. |
| Gemma 4 31B30.7B | Q8_0 | 33 GB | ~45 tok/s ModelFit est. |
| Qwen3-Coder-30B-A3B30.5B MoE | Q8_0 | 34 GB | ~193 tok/s ModelFit est. |
| Nemotron 3 Nano 30B-A3B30B MoE | Q8_0 | 34 GB | ~309 tok/s our est. |
| Qwen3.8-27B27.3B | Q8_0 | 29 GB | ~54 tok/s ModelFit est. |
| Gemma 4 26B-A4B25.2B MoE | Q8_0 | 30 GB | ~184 tok/s ModelFit est. |
| gpt-oss-20b21B MoE | MXFP4 | 15.1 GB | ~242 tok/s ModelFit est. |
| Nemotron Nano 9B v29B | FP16 | 19.4 GB | ~55 tok/s our est. |
| Qwen3.5 9B Instruct9B | Q8_0 | 12.0 GB | ~146 tok/s ModelFit est. |
| Qwen2.5 7B Instruct7B | Q4_K_M | 6.6 GB | ~224 tok/s ModelFit est. |
| Gemma 3 4B Instruct4B | Q8_0 | 5.0 GB | ~232 tok/s ModelFit est. |
| Qwen3-Next 80B-A3B80B MoE | Q4_K_M | 53 GB | ~285 tok/s ModelFit est. |
| Kolibri 178.1B | Q4_K_M | 52 GB | ~260 tok/s ModelFit est. |
| Llama 3.1 70B Instruct70B | Q4_K_M | 44 GB | ~36 tok/s ModelFit est. |
| DeepSeek-R1 Distill Llama 70B70B | Q4_K_M | 44 GB | ~36 tok/s ModelFit est. |
| Mixtral 8x7B Instruct46.7B MoE | Q4_K_M | 32 GB | ~108 tok/s ModelFit est. |
| Flint 40B-A2.5B39.8B MoE | NVFP4 | 24 GB | ~327 tok/s ModelFit est. |
| Sage-A1.535.95B | Q4_K_M | 23 GB | ~314 tok/s ModelFit est. |
| Qwen3.5 35B-A3B Instruct35B MoE | Q8_0 | 41 GB | ~234 tok/s ModelFit est. |
| Ornith 1.0 35B35B | Q4_K_M | 23 GB | ~317 tok/s ModelFit est. |
| Nanbeige4.3 33B33.1B | Q8_0 | 37 GB | ~44 tok/s ModelFit est. |
| Poolside Laguna XS.233B | Q4_K_M | 25 GB | no estimate |
| Laguna XS 2.133B | Q4_K_M | 22 GB | ~302 tok/s ModelFit est. |
| Qwen3 32B32B | Q8_0 | 37 GB | ~43 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 32B32B | Q8_0 | 37 GB | ~43 tok/s ModelFit est. |
| Olmo 3.1 32B Instruct32B | Q4_K_M | 21 GB | ~71 tok/s ModelFit est. |
| Qwen3 30B30B | Q8_0 | 32 GB | ~193 tok/s ModelFit est. |
| NVIDIA Nemotron Cascade 2 30B-A3B30B MoE | Q6_K | 26 GB | ~411 tok/s our est. |
| Cohere North Mini Code30B | Q4_K_M | 20 GB | no estimate |
| Nemotron 3.5 Lightning 30B-A3B30B MoE | Q4_K_M | 27 GB | ~233 tok/s ModelFit est. |
| GLM-4.7-Flash 30B-A3B30B MoE | Q4_K_M | 20 GB | ~252 tok/s ModelFit est. |
| Granite Switch 4.1 30B30B | Q4_K_M | 23 GB | ~66 tok/s ModelFit est. |
| Muse Glimmer 30B29.8B | Q4_K_M | 18.3 GB | ~88 tok/s ModelFit est. |
| Granite 4.2 30B29.3B | Q4_K_M | 17.9 GB | ~77 tok/s ModelFit est. |
| AREX-2 27B27.4B | Q4_K_M | 18.6 GB | ~85 tok/s ModelFit est. |
| Gemma 2 27B Instruct27B | Q4_K_M | 23 GB | ~78 tok/s ModelFit est. |
| Gemma 3 27B Instruct27B | Q4_K_M | 23 GB | ~82 tok/s ModelFit est. |
| Qwen3.5 27B Instruct27B | Q8_0 | 29 GB | ~54 tok/s ModelFit est. |
| Qwen3.6 27B27B | Q8_0 | 32 GB | ~54 tok/s ModelFit est. |
| Jebadiah 27B27B | Q8_0 | 31 GB | ~54 tok/s ModelFit est. |
| Mistral Small 3.124B | Q8_0 | 25 GB | ~57 tok/s ModelFit est. |
| LFM2 24B-A2B Instruct24B MoE | Q4_K_M | 15.3 GB | ~330 tok/s ModelFit est. |
| Devstral Small 2 24B24B | Q4_K_M | 16.6 GB | ~90 tok/s ModelFit est. |
| Mistral Small 22B22B | Q4_K_M | 18.4 GB | ~102 tok/s ModelFit est. |
| Qwen2.5 14B Instruct14B | Q4_K_M | 12.3 GB | ~136 tok/s ModelFit est. |
| Phi-3 Medium 14B14B | Q4_K_M | 12.3 GB | ~90 tok/s our est. |
| Phi-4 14B14B | Q8_0 | 15.9 GB | ~88 tok/s ModelFit est. |
| Qwen2.5 Coder 14B14B | Q4_K_M | 12.3 GB | ~136 tok/s ModelFit est. |
| Qwen3 14B14B | Q8_0 | 17.3 GB | ~89 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 14B14B | Q8_0 | 16.0 GB | ~89 tok/s ModelFit est. |
| Ministral 3 14B13.5B | Q4_K_M | 10.3 GB | ~136 tok/s ModelFit est. |
| Mellum2.1 12B-A2.5B Thinking12.1B MoE | Q4_K_M | 9.3 GB | ~589 tok/s our est. |
| Mistral Nemo 12B12B | Q8_0 | 13.4 GB | ~104 tok/s ModelFit est. |
| Gemma 3 12B Instruct12B | Q4_K_M | 10.7 GB | ~154 tok/s ModelFit est. |
| Gemma 4 12B12B | Q8_0 | 14.1 GB | ~103 tok/s ModelFit est. |
| GLM-4-9B-Chat9.4B | Q8_0 | 11.2 GB | ~134 tok/s ModelFit est. |
| Gemma 2 9B Instruct9B | Q4_K_M | 8.2 GB | ~165 tok/s ModelFit est. |
| Ornith 1.0 9B9B | Q4_K_M | 6.7 GB | ~211 tok/s ModelFit est. |
| MiMo-V2.6-Distill-Qwen-9B9B | Q4_K_M | 6.9 GB | ~206 tok/s ModelFit est. |
| Jebadiah 9B9B | Q8_0 | 11.0 GB | ~144 tok/s ModelFit est. |
| Granite 4.2 8B8.8B | Q4_K_M | 6.1 GB | ~193 tok/s ModelFit est. |
| Granite Switch 4.1 8B8.8B | Q4_K_M | 7.6 GB | ~166 tok/s ModelFit est. |
| MiniCPM-V 4.5 8B8.7B | Q4_K_M | 6.8 GB | ~205 tok/s ModelFit est. |
| Ministral 3 8B8.4B | Q4_K_M | 7.1 GB | ~183 tok/s ModelFit est. |
| LFM2.5 8B-A1B8.3B MoE | Q4_K_M | 6.6 GB | ~374 tok/s ModelFit est. |
| Llama 3.1 8B Instruct8B | Q8_0 | 9.2 GB | ~146 tok/s ModelFit est. |
| Qwen3 8B8B | Q8_0 | 9.3 GB | ~142 tok/s ModelFit est. |
| Granite 4.1 8B Instruct8B | Q4_K_M | 6.6 GB | ~187 tok/s ModelFit est. |
| Mistral 7B Instruct7B | Q8_0 | 8.4 GB | ~151 tok/s ModelFit est. |
| Qwen2.5 Coder 7B7B | Q4_K_M | 6.6 GB | ~224 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 7B7B | Q8_0 | 8.7 GB | ~157 tok/s ModelFit est. |
| Gemma 4 E4B4.5B | Q8_0 | 8.7 GB | ~211 tok/s ModelFit est. |
| Qwen3.5 4B Instruct4B | Q8_0 | 5.4 GB | ~223 tok/s ModelFit est. |
| NVIDIA Nemotron 3.5 Bolt E4B4B | Q4_K_M | 9.1 GB | ~248 tok/s ModelFit est. |
| Phi-3 Mini 3.8B3.8B | Q4_K_M | 4.3 GB | ~308 tok/s our est. |
| Phi-4 Mini 3.8B3.8B | Q8_0 | 4.9 GB | ~223 tok/s ModelFit est. |
| Granite 4.2 3B3.7B | Q4_K_M | 3.2 GB | ~318 tok/s ModelFit est. |
| Ministral 3 3B3.4B | Q4_K_M | 4.1 GB | ~268 tok/s ModelFit est. |
| Qwen2.5 3B Instruct3B | Q4_K_M | 3.6 GB | ~341 tok/s ModelFit est. |
| Llama 3.2 3B Instruct3B | Q4_K_M | 3.6 GB | ~314 tok/s ModelFit est. |
| Granite 4.1 3B Instruct3B | Q4_K_M | 3.0 GB | ~317 tok/s ModelFit est. |
| Gemma 4 E2B2.3B | Q8_0 | 5.7 GB | ~341 tok/s ModelFit est. |
| Gemma 2 2B Instruct2B | Q4_K_M | 2.8 GB | ~336 tok/s ModelFit est. |
| Qwen3.5 2B Instruct2B | Q4_K_M | 2.8 GB | ~343 tok/s ModelFit est. |
| LFM3 1.7B1.73B | Q4_K_M | 2.1 GB | ~423 tok/s ModelFit est. |
| Qwen2.5 1.5B Instruct1.5B | Q4_K_M | 2.5 GB | ~428 tok/s ModelFit est. |
| Llama 3.2 1B Instruct1B | Q4_K_M | 2.0 GB | ~447 tok/s ModelFit est. |
| Gemma 3 1B Instruct1B | Q8_0 | 2.0 GB | ~429 tok/s ModelFit est. |
| Qwen3.5 0.8B Instruct0.8B | Q4_K_M | 1.8 GB | ~457 tok/s ModelFit est. |
| Qwen2.5 0.5B Instruct0.5B | Q4_K_M | 1.8 GB | ~517 tok/s ModelFit est. |
| SmolLM2 360M0.36B | Q4_K_M | 1.5 GB | ~453 tok/s ModelFit est. |
Speeds come from ModelFit's estimates (CC BY 4.0) where it covers the model, otherwise from our bandwidth formula. Method
Fits only at 3-bit or lower
Too big for this machine
Nemotron 3 Super 120B-A12B (67 GB), gpt-oss-120b (68 GB), Llama 4 Scout (70 GB), Ling-3.0-flash (71 GB), Qwen3.5 122B-A10B Instruct (75 GB), Devstral 2 123B (78 GB), Ornith 1.5 122B (79 GB), Qwen3.8-Flash-Next (93 GB), Laguna S 2.1 (99 GB), Qwen3-235B-A22B (134 GB), DeepSeek-V4-Flash (177 GB), GLM-5.3-Flash (200 GB) and more.
Questions
What is the best AI model for the 2x RTX 5090 (64 GB)?
Llama 3.3 70B at Q6_K: the largest model on our shortlist of current models that runs at 4-bit or better at a usable speed (about 17 tok/s, our est.).
How much memory can a model use on the 2x RTX 5090 (64 GB)?
About 62 GB of the 64 GB. Two cards in one PC, model split across both. Each token still reads at one card's bandwidth.
How fast is the 2x RTX 5090 (64 GB) for local LLMs?
Decode speed is limited by memory bandwidth: 1792 GB/s. A model that reads 10 GB per token runs at under 179 tokens per second.