NVIDIA graphics card
What AI models can the NVIDIA GeForce RTX 4080 SUPER 16 GB run?
At 4-bit quantization with an 8K-token conversation, 28 of the 146 open-weight models in our directory fit entirely in the NVIDIA GeForce RTX 4080 SUPER 16 GB's 16 GB of video memory (about 15 GB usable), up to Safety GPT OSS 20B. With 32 GB of dual-channel DDR5-5600, 33 more run partly from system RAM, more slowly. Among models of 7B parameters or more, GPT OSS 20B is the fastest, at an estimated 170 to 220 tokens per second.
| Model | Parameters | Memory needed | Estimated speed |
|---|---|---|---|
| Runs entirely in fast memory (28) | |||
| Safety GPT OSS 20B OpenAI | 21.5B 4.8B active | 14 GB 13 GB + 0.2 GB cache | 153–199 tokens/s Fast |
| GPT OSS 20B OpenAI | 20.9B 4.2B active | 14 GB 13 GB + 0.2 GB cache | 170–220 tokens/s Fast |
| Reka Flash 3 Rekaai | 20.9B | 15 GB 13 GB + 1.1 GB cache | 41–52 tokens/s Fast, little memory to spare |
| Qwen3 14B Alibaba (Qwen) | 14.8B | 11 GB 9.0 GB + 1.3 GB cache | 55–71 tokens/s Fast |
| Phi 4 Microsoft | 14.7B | 11 GB 8.9 GB + 1.7 GB cache | 54–70 tokens/s Fast |
| Ministral 3 14B Mistral AI | 13.9B | 11 GB 8.5 GB + 1.3 GB cache | 58–74 tokens/s Fast |
| Gemma 3 12B IT Google DeepMind | 12.2B | 8.2 GB weights only | Speed not estimated (architecture not public) |
| Llama Guard 4 12B Meta | 12.0B | 8.1 GB weights only | Speed not estimated (architecture not public) |
| Qwen3.5 9B Alibaba (Qwen) | 9.7B | 6.9 GB 5.9 GB + 0.3 GB cache | 85–110 tokens/s Fast |
| Moondream 3.1 Moondream | 9.3B | 6.4 GB weights only | Speed not estimated (architecture not public) |
| Moondream 3 Preview Moondream | 9.3B | 6.4 GB weights only | Speed not estimated (architecture not public) |
| Ministral 3 8B Mistral AI | 8.9B | 7.3 GB 5.4 GB + 1.1 GB cache | 85–110 tokens/s Fast |
| Granite 4.2 8B IBM | 8.8B | 7.5 GB 5.3 GB + 1.3 GB cache | 84–109 tokens/s Fast |
| Qwen3 VL 8B Instruct Alibaba (Qwen) | 8.8B | 7.3 GB 5.3 GB + 1.2 GB cache | 86–111 tokens/s Fast |
| Qwen3 VL 8B Thinking Alibaba (Qwen) | 8.8B | 7.3 GB 5.3 GB + 1.2 GB cache | 86–111 tokens/s Fast |
| UI-TARS 7B Bytedance | 8.3B | 6.3 GB 5.0 GB + 0.5 GB cache | 95–123 tokens/s Fast |
| Qwen3 8B Alibaba (Qwen) | 8.2B | 7.0 GB 5.0 GB + 1.2 GB cache | 90–117 tokens/s Fast |
| Llama 3.1 8B Instruct fp8 Meta | 8.0B | 5.7 GB weights only | Speed not estimated (architecture not public) |
| Hy-MT2-7B Tencent (Hunyuan) | 8.0B | 6.7 GB 4.9 GB + 1.1 GB cache | 93–120 tokens/s Fast |
| Qwen2.5 7B Instruct Alibaba (Qwen) | 7.6B | 5.9 GB 4.6 GB + 0.5 GB cache | 102–132 tokens/s Fast |
| Gemma 3 4B IT Google DeepMind | 4.3B | 3.4 GB weights only | Speed not estimated (architecture not public) |
| Nemotron 3.5 Content Safety (free) NVIDIA | 4.3B | 3.7 GB 2.6 GB + 0.3 GB cache | 165–213 tokens/s Fast |
| Ministral 3 3B Mistral AI | 3.8B | 4.0 GB 2.3 GB + 0.9 GB cache | 164–212 tokens/s Fast |
| Llama 3.2 3B Instruct Meta | 3.2B | 2.7 GB weights only | Speed not estimated (architecture not public) |
| Granite 4.0 Micro IBM | 3.2B | 2.8 GB 1.9 GB + 0.1 GB cache | 214–277 tokens/s Fast |
| Hy-MT2-1.8B Tencent (Hunyuan) | 2.0B | 2.6 GB 1.2 GB + 0.5 GB cache | 260–336 tokens/s Fast |
| Hy-MT2-1.8B Tencent (Hunyuan) | 2.0B | 2.6 GB 1.2 GB + 0.5 GB cache | 260–336 tokens/s Fast |
| Llama 3.2 1B Instruct Meta | 1.2B | 1.5 GB weights only | Speed not estimated (architecture not public) |
| Runs, partly from system RAM (slower) (33) | |||
| Skyfall 36B V2 Thedrummer | 36.9B | 25 GB 22 GB + 2.1 GB cache | 2.9–3.9 tokens/s Too slow for chat, 9.9 GB in system RAM |
| Qwen3.5 35B-A3B Alibaba (Qwen) | 36.0B 4.7B active | 23 GB 22 GB + 0.2 GB cache | 65–90 tokens/s Fast, 7.4 GB in system RAM |
| Qwen3.6 35B-A3B Alibaba (Qwen) | 36.0B 4.7B active | 23 GB 22 GB + 0.2 GB cache | 65–90 tokens/s Fast, 7.4 GB in system RAM |
| Nex-N2.5-Mini NEX AGI | 35.1B 3.9B active | 22 GB 21 GB + 0.2 GB cache | 73–101 tokens/s Fast, 6.9 GB in system RAM |
| Laguna XS 2.1 (free) Poolside | 33.4B 2.2B active | 21 GB 20 GB + 0.4 GB cache | 93–127 tokens/s Fast, 6.1 GB in system RAM |
| Qwen3 VL 32B Instruct Alibaba (Qwen) | 33.4B | 23 GB 20 GB + 2.1 GB cache | 3.6–4.9 tokens/s Usable but slow, 7.8 GB in system RAM |
| Qwen2.5 Coder 32B Instruct Alibaba (Qwen) | 32.8B | 23 GB 20 GB + 2.1 GB cache | 3.7–5.1 tokens/s Usable but slow, 7.4 GB in system RAM |
| Qwen3 32B Alibaba (Qwen) | 32.8B | 23 GB 20 GB + 2.1 GB cache | 3.7–5.1 tokens/s Usable but slow, 7.4 GB in system RAM |
| Nemotron 3 Nano 30B A3B NVIDIA | 31.6B | 20 GB 19 GB + 0.1 GB cache | 4.6 GB in system RAM; Speed not estimated for this architecture |
| Nemotron 3.5 Lightning 30B A3B NVIDIA | 31.6B | 20 GB 19 GB + 0.4 GB cache | 5.0 GB in system RAM; Speed not estimated for this architecture |
| Gemma 4 31B IT Google DeepMind | 31.3B | 22 GB 19 GB + 2.2 GB cache | 4.2–5.7 tokens/s Usable but slow, 6.5 GB in system RAM |
| GLM-4.7-Flash Zhipu AI (Z.ai) | 31.2B 5.2B active | 20 GB 19 GB + 0.4 GB cache | 55–75 tokens/s Fast, 4.8 GB in system RAM |
| Qwen3 VL 30B A3B Instruct Alibaba (Qwen) | 31.1B 3.9B active | 20 GB 19 GB + 0.8 GB cache | 58–79 tokens/s Fast, 5.0 GB in system RAM |
| Qwen3 VL 30B A3B Thinking Alibaba (Qwen) | 31.1B 3.9B active | 20 GB 19 GB + 0.8 GB cache | 58–79 tokens/s Fast, 5.0 GB in system RAM |
| Qwen3-Coder 30B-A3B Instruct Alibaba (Qwen) | 30.5B 3.4B active | 20 GB 19 GB + 0.8 GB cache | 63–86 tokens/s Fast, 4.7 GB in system RAM |
| Qwen3 30B A3B Instruct 2507 Alibaba (Qwen) | 30.5B 3.4B active | 20 GB 19 GB + 0.8 GB cache | 63–86 tokens/s Fast, 4.7 GB in system RAM |
| Qwen3 30B A3B Thinking 2507 Alibaba (Qwen) | 30.5B 3.4B active | 20 GB 19 GB + 0.8 GB cache | 63–86 tokens/s Fast, 4.7 GB in system RAM |
| Qwen3 30B A3B Alibaba (Qwen) | 30.5B 3.4B active | 20 GB 19 GB + 0.8 GB cache | 63–86 tokens/s Fast, 4.7 GB in system RAM |
| Hy-MT2-30B-A3B Tencent (Hunyuan) | 30.1B 3.5B active | 20 GB 18 GB + 0.8 GB cache | 64–88 tokens/s Fast, 4.4 GB in system RAM |
| Muse Glimmer 30B Meta | 29.8B | 19 GB 18 GB + 0.2 GB cache | 6.6–9.1 tokens/s Usable but slow, 3.6 GB in system RAM |
| Qwen3.8 27B Alibaba (Qwen) | 27.8B | 18 GB 17 GB + 0.5 GB cache | 8.1–11 tokens/s Usable but slow, 2.8 GB in system RAM |
| Qwen3.5 27B Alibaba (Qwen) | 27.8B | 18 GB 17 GB + 0.5 GB cache | 8.1–11 tokens/s Usable but slow, 2.8 GB in system RAM |
| Qwen3.6 27B Alibaba (Qwen) | 27.8B | 18 GB 17 GB + 0.5 GB cache | 8.1–11 tokens/s Usable but slow, 2.8 GB in system RAM |
| Gemma 3 27B IT Google DeepMind | 27.4B | 17 GB weights only | 2.0 GB in system RAM; Speed not estimated (architecture not public) |
| Gemma 2 27B Google DeepMind | 27.2B | 17 GB weights only | 1.9 GB in system RAM; Speed not estimated (architecture not public) |
| Gemma 4 26B A4B IT Google DeepMind | 25.8B | 17 GB 16 GB + 0.5 GB cache | 1.6 GB in system RAM; Speed not estimated for this architecture |
| Voxtral Small 24B 2507 Mistral AI | 24.3B | 17 GB 15 GB + 1.3 GB cache | 12–17 tokens/s Comfortable to read, 1.5 GB in system RAM |
| Uncensored Cognitivecomputations | 24.0B | 17 GB 15 GB + 1.3 GB cache | 13–18 tokens/s Comfortable to read, 1.3 GB in system RAM |
| Mistral Small 3.1 24B Instruct Mistral AI | 24.0B | 17 GB 15 GB + 1.3 GB cache | 13–18 tokens/s Comfortable to read, 1.3 GB in system RAM |
| LFM2.5-24B-A2B Liquid | 24.0B | 16 GB 15 GB + 0.2 GB cache | 0.1 GB in system RAM; Speed not estimated for this architecture |
| Mistral Small 3 Mistral AI | 23.6B | 16 GB 14 GB + 1.3 GB cache | 15–20 tokens/s Comfortable to read, 1.0 GB in system RAM |
| Cydonia 24B V4.1 Thedrummer | 23.6B | 16 GB 14 GB + 1.3 GB cache | 15–20 tokens/s Comfortable to read, 1.0 GB in system RAM |
| Cydonia 24B V4.1 Thedrummer | 23.6B | 16 GB 14 GB + 1.3 GB cache | 15–20 tokens/s Comfortable to read, 1.0 GB in system RAM |
85 larger models do not fit. Change the quantization, conversation length or system RAM in the tool.
Compared with similar cards
- NVIDIA GeForce RTX 5070 12 GB: 12 GB, 25 models fit entirely (-3); about 8% slower on models that fit both (672 vs 736 GB/s).
- NVIDIA GeForce RTX 5080 16 GB: same memory, so the same models fit; about 26% faster on models that fit both (960 vs 736 GB/s).
- NVIDIA GeForce RTX 5070 Ti 16 GB: same memory, so the same models fit; about 19% faster on models that fit both (896 vs 736 GB/s).
- NVIDIA GeForce RTX 4070 Ti SUPER 16 GB: same memory, so the same models fit; about 8% slower on models that fit both (672 vs 736 GB/s).
- NVIDIA GeForce RTX 4090 24 GB: 24 GB, 60 models fit entirely (+32); about 31% faster on models that fit both (1008 vs 736 GB/s).
Frequently asked questions
What is the largest AI model the NVIDIA GeForce RTX 4080 SUPER 16 GB can run?
At 4-bit quantization with an 8K-token conversation, the largest model in our directory that fits entirely in its 16 GB of video memory is Safety GPT OSS 20B (14 GB). With 32 GB of dual-channel DDR5-5600 for the overflow, larger models such as Skyfall 36B V2 also run, more slowly, partly from system RAM.
Can the NVIDIA GeForce RTX 4080 SUPER 16 GB run GPT OSS 20B?
Yes. At 4-bit, GPT OSS 20B needs about 14 GB and fits entirely in the NVIDIA GeForce RTX 4080 SUPER 16 GB's memory at an estimated 170 to 220 tokens per second.
Can the NVIDIA GeForce RTX 4080 SUPER 16 GB run Qwen3 32B?
Partly. Qwen3 32B needs about 23 GB at 4-bit, so about 7.4 GB runs from system RAM (assuming 32 GB of dual-channel DDR5-5600) at an estimated 3.7 to 5.1 tokens per second.
Can the NVIDIA GeForce RTX 4080 SUPER 16 GB run OpenAI GPT 120B?
No. OpenAI GPT 120B needs about 72 GB at 4-bit, more than the NVIDIA GeForce RTX 4080 SUPER 16 GB can hold even with 32 GB of dual-channel DDR5-5600 for overflow.
These are estimates, not measurements: memory from each model's parameter count and published architecture, speed from memory bandwidth, calibrated on the AIPCs test bench. Drivers, software and prompt length move real results. How the estimate works.