Graphics cards
Which AI models run on the 2x RTX 4090 (48 GB)?
The 2x RTX 4090 (48 GB) has 48 GB of memory, about 46 GB of it usable for a model, and 1008 GB/s of memory bandwidth. 92 of the 110 models in our catalog run on it at 4-bit or better.
Data as of 2 November 2026
48 GBmemory
~46 GB usable for a model
1008 GB/sbandwidth, sets the speed
92models 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 | Q4_K_M | 44 GB | ~13 tok/s our est. |
| Llama 3.3 Nemotron Super 49B49B | Q6_K | 42 GB | ~14 tok/s our est. |
| Qwen3.6-35B-A3B35B MoE | Q8_0 | 41 GB | ~154 tok/s ModelFit est. |
| Gemma 4 31B30.7B | Q8_0 | 33 GB | ~25 tok/s ModelFit est. |
| Qwen3-Coder-30B-A3B30.5B MoE | Q8_0 | 34 GB | ~122 tok/s ModelFit est. |
| Nemotron 3 Nano 30B-A3B30B MoE | Q8_0 | 34 GB | ~174 tok/s our est. |
| Qwen3.8-27B27.3B | Q8_0 | 29 GB | ~30 tok/s ModelFit est. |
| Gemma 4 26B-A4B25.2B MoE | Q8_0 | 30 GB | ~116 tok/s ModelFit est. |
| gpt-oss-20b21B MoE | MXFP4 | 15.1 GB | ~161 tok/s ModelFit est. |
| Nemotron Nano 9B v29B | FP16 | 19.4 GB | ~31 tok/s our est. |
| Qwen3.5 9B Instruct9B | Q8_0 | 12.0 GB | ~89 tok/s ModelFit est. |
| Qwen2.5 7B Instruct7B | Q4_K_M | 6.6 GB | ~146 tok/s ModelFit est. |
| Gemma 3 4B Instruct4B | Q8_0 | 5.0 GB | ~152 tok/s ModelFit est. |
| Kolibri 178.1B | IQ4_XS | 46 GB | no estimate |
| Llama 3.1 70B Instruct70B | Q4_K_M | 44 GB | ~13 tok/s our est. |
| DeepSeek-R1 Distill Llama 70B70B | Q4_K_M | 44 GB | ~13 tok/s our est. |
| Mixtral 8x7B Instruct46.7B MoE | Q4_K_M | 32 GB | ~63 tok/s ModelFit est. |
| Flint 40B-A2.5B39.8B MoE | NVFP4 | 24 GB | ~236 tok/s ModelFit est. |
| Sage-A1.535.95B | Q4_K_M | 23 GB | ~223 tok/s ModelFit est. |
| Qwen3.5 35B-A3B Instruct35B MoE | Q8_0 | 41 GB | ~154 tok/s ModelFit est. |
| Ornith 1.0 35B35B | Q4_K_M | 23 GB | ~226 tok/s ModelFit est. |
| Nanbeige4.3 33B33.1B | Q8_0 | 37 GB | ~24 tok/s ModelFit est. |
| Poolside Laguna XS.233B | Q4_K_M | 25 GB | no estimate |
| Laguna XS 2.133B | Q4_K_M | 22 GB | ~212 tok/s ModelFit est. |
| Qwen3 32B32B | Q8_0 | 37 GB | ~24 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 32B32B | Q8_0 | 37 GB | ~24 tok/s ModelFit est. |
| Olmo 3.1 32B Instruct32B | Q4_K_M | 21 GB | ~41 tok/s ModelFit est. |
| Qwen3 30B30B | Q8_0 | 32 GB | ~122 tok/s ModelFit est. |
| NVIDIA Nemotron Cascade 2 30B-A3B30B MoE | Q6_K | 26 GB | ~231 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 | ~153 tok/s ModelFit est. |
| GLM-4.7-Flash 30B-A3B30B MoE | Q4_K_M | 20 GB | ~169 tok/s ModelFit est. |
| Granite Switch 4.1 30B30B | Q4_K_M | 23 GB | ~38 tok/s ModelFit est. |
| Muse Glimmer 30B29.8B | Q4_K_M | 18.3 GB | ~51 tok/s ModelFit est. |
| Granite 4.2 30B29.3B | Q4_K_M | 17.9 GB | ~44 tok/s ModelFit est. |
| AREX-2 27B27.4B | Q4_K_M | 18.6 GB | ~49 tok/s ModelFit est. |
| Gemma 2 27B Instruct27B | Q4_K_M | 23 GB | ~45 tok/s ModelFit est. |
| Gemma 3 27B Instruct27B | Q4_K_M | 23 GB | ~47 tok/s ModelFit est. |
| Qwen3.5 27B Instruct27B | Q8_0 | 29 GB | ~30 tok/s ModelFit est. |
| Qwen3.6 27B27B | Q8_0 | 32 GB | ~30 tok/s ModelFit est. |
| Jebadiah 27B27B | Q8_0 | 31 GB | ~30 tok/s ModelFit est. |
| Mistral Small 3.124B | Q8_0 | 25 GB | ~32 tok/s ModelFit est. |
| LFM2 24B-A2B Instruct24B MoE | Q4_K_M | 15.3 GB | ~239 tok/s ModelFit est. |
| Devstral Small 2 24B24B | Q4_K_M | 16.6 GB | ~52 tok/s ModelFit est. |
| Mistral Small 22B22B | Q4_K_M | 18.4 GB | ~60 tok/s ModelFit est. |
| Qwen2.5 14B Instruct14B | Q4_K_M | 12.3 GB | ~82 tok/s ModelFit est. |
| Phi-3 Medium 14B14B | Q4_K_M | 12.3 GB | ~50 tok/s our est. |
| Phi-4 14B14B | Q8_0 | 15.9 GB | ~51 tok/s ModelFit est. |
| Qwen2.5 Coder 14B14B | Q4_K_M | 12.3 GB | ~82 tok/s ModelFit est. |
| Qwen3 14B14B | Q8_0 | 17.3 GB | ~52 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 14B14B | Q8_0 | 16.0 GB | ~52 tok/s ModelFit est. |
| Ministral 3 14B13.5B | Q4_K_M | 10.3 GB | ~82 tok/s ModelFit est. |
| Mellum2.1 12B-A2.5B Thinking12.1B MoE | Q4_K_M | 9.3 GB | ~331 tok/s our est. |
| Mistral Nemo 12B12B | Q8_0 | 13.4 GB | ~61 tok/s ModelFit est. |
| Gemma 3 12B Instruct12B | Q4_K_M | 10.7 GB | ~94 tok/s ModelFit est. |
| Gemma 4 12B12B | Q8_0 | 14.1 GB | ~60 tok/s ModelFit est. |
| GLM-4-9B-Chat9.4B | Q8_0 | 11.2 GB | ~81 tok/s ModelFit est. |
| Gemma 2 9B Instruct9B | Q4_K_M | 8.2 GB | ~102 tok/s ModelFit est. |
| Ornith 1.0 9B9B | Q4_K_M | 6.7 GB | ~136 tok/s ModelFit est. |
| MiMo-V2.6-Distill-Qwen-9B9B | Q4_K_M | 6.9 GB | ~132 tok/s ModelFit est. |
| Jebadiah 9B9B | Q8_0 | 11.0 GB | ~87 tok/s ModelFit est. |
| Granite 4.2 8B8.8B | Q4_K_M | 6.1 GB | ~122 tok/s ModelFit est. |
| Granite Switch 4.1 8B8.8B | Q4_K_M | 7.6 GB | ~103 tok/s ModelFit est. |
| MiniCPM-V 4.5 8B8.7B | Q4_K_M | 6.8 GB | ~132 tok/s ModelFit est. |
| Ministral 3 8B8.4B | Q4_K_M | 7.1 GB | ~115 tok/s ModelFit est. |
| LFM2.5 8B-A1B8.3B MoE | Q4_K_M | 6.6 GB | ~283 tok/s ModelFit est. |
| Llama 3.1 8B Instruct8B | Q8_0 | 9.2 GB | ~89 tok/s ModelFit est. |
| Qwen3 8B8B | Q8_0 | 9.3 GB | ~86 tok/s ModelFit est. |
| Granite 4.1 8B Instruct8B | Q4_K_M | 6.6 GB | ~118 tok/s ModelFit est. |
| Mistral 7B Instruct7B | Q8_0 | 8.4 GB | ~93 tok/s ModelFit est. |
| Qwen2.5 Coder 7B7B | Q4_K_M | 6.6 GB | ~146 tok/s ModelFit est. |
| DeepSeek-R1 Distill Qwen 7B7B | Q8_0 | 8.7 GB | ~96 tok/s ModelFit est. |
| Gemma 4 E4B4.5B | Q8_0 | 8.7 GB | ~136 tok/s ModelFit est. |
| Qwen3.5 4B Instruct4B | Q8_0 | 5.4 GB | ~146 tok/s ModelFit est. |
| NVIDIA Nemotron 3.5 Bolt E4B4B | Q4_K_M | 9.1 GB | ~165 tok/s ModelFit est. |
| Phi-3 Mini 3.8B3.8B | Q4_K_M | 4.3 GB | ~173 tok/s our est. |
| Phi-4 Mini 3.8B3.8B | Q8_0 | 4.9 GB | ~145 tok/s ModelFit est. |
| Granite 4.2 3B3.7B | Q4_K_M | 3.2 GB | ~227 tok/s ModelFit est. |
| Ministral 3 3B3.4B | Q4_K_M | 4.1 GB | ~182 tok/s ModelFit est. |
| Qwen2.5 3B Instruct3B | Q4_K_M | 3.6 GB | ~249 tok/s ModelFit est. |
| Llama 3.2 3B Instruct3B | Q4_K_M | 3.6 GB | ~223 tok/s ModelFit est. |
| Granite 4.1 3B Instruct3B | Q4_K_M | 3.0 GB | ~226 tok/s ModelFit est. |
| Gemma 4 E2B2.3B | Q8_0 | 5.7 GB | ~249 tok/s ModelFit est. |
| Gemma 2 2B Instruct2B | Q4_K_M | 2.8 GB | ~244 tok/s ModelFit est. |
| Qwen3.5 2B Instruct2B | Q4_K_M | 2.8 GB | ~251 tok/s ModelFit est. |
| LFM3 1.7B1.73B | Q4_K_M | 2.1 GB | ~337 tok/s ModelFit est. |
| Qwen2.5 1.5B Instruct1.5B | Q4_K_M | 2.5 GB | ~344 tok/s ModelFit est. |
| Llama 3.2 1B Instruct1B | Q4_K_M | 2.0 GB | ~366 tok/s ModelFit est. |
| Gemma 3 1B Instruct1B | Q8_0 | 2.0 GB | ~345 tok/s ModelFit est. |
| Qwen3.5 0.8B Instruct0.8B | Q4_K_M | 1.8 GB | ~379 tok/s ModelFit est. |
| Qwen2.5 0.5B Instruct0.5B | Q4_K_M | 1.8 GB | ~461 tok/s ModelFit est. |
| SmolLM2 360M0.36B | Q4_K_M | 1.5 GB | ~374 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
Too big for this machine
Qwen3-Next 80B-A3B (53 GB), 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), Nemotron 3.5 Super VL 120B-A12B (76 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) and more.
Questions
What is the best AI model for the 2x RTX 4090 (48 GB)?
Llama 3.3 70B at Q4_K_M: the largest model on our shortlist of current models that runs at 4-bit or better at a usable speed (about 13 tok/s, our est.).
How much memory can a model use on the 2x RTX 4090 (48 GB)?
About 46 GB of the 48 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 4090 (48 GB) for local LLMs?
Decode speed is limited by memory bandwidth: 1008 GB/s. A model that reads 10 GB per token runs at under 101 tokens per second.