GPUs & Macs › NVIDIA RTX 2080 Ti
NVIDIA RTX 2080 Ti: local LLM speed
Llama 2 7B (Q4_0) generates 109 tokens/s on the NVIDIA RTX 2080 Ti (CUDA), and processes prompts at 3,108 tokens/s.
Measured results
| Model | Generate tok/s | Prompt tok/s | Quant | Backend | Test | Source |
|---|---|---|---|---|---|---|
| Llama 2 7B | 109 | 3,108 | Q4_0 | CUDA FA on | tg128 | llama.cpp #15013 details |
| Llama 2 7B | 108 | 2,891 | Q4_0 | CUDA FA off | tg128 | llama.cpp #15013 details |
| Llama 2 7B | 101 | 1,936 | Q4_0 | Vulkan FA on | tg128 | llama.cpp #10879 details |
| Llama 2 7B | 97.6 | 1,888 | Q4_0 | Vulkan FA off | tg128 | llama.cpp #10879 details |
| Llama 3.2 1B Instruct | 222 | 8,722 | Q4_K_M | llamafile | LocalScore avg | LocalScore details |
| Llama 3.1 8B Instruct | 72.9 | 2,027 | Q4_K_M | llamafile | LocalScore avg | LocalScore details |
| Qwen2.5 14B Instruct | 39.5 | 1,140 | Q4_K_M | llamafile | LocalScore avg | LocalScore details |
What fits on the NVIDIA RTX 2080 Ti
Estimates for this chip are scaled from a measured run: Llama 2 7B Q4_0 at 109 tok/s (CUDA, source). Fit assumes the whole model in GPU memory with ~8K context.
| Model | Best quant that fits | File | Expected tok/s |
|---|---|---|---|
| Qwen3.8-27B | UD-IQ2_XXS | 7.3 GB | 57.4 estimate |
| gemma-4-12B-it | Q5_K_M | 8.4 GB | 49.6 estimate |
| Qwen3.5-9B | Q6_K | 7.5 GB | 56.0 estimate |
| Qwen3.5-4B | Q8_0 | 4.5 GB | 93.1 estimate |
| MiMo-V2.6-Distill-Qwen-9B | Q6_K | 7.8 GB | 53.5 estimate |
| gemma-4-E4B-it | Q8_0 | 8.2 GB | 50.9 estimate |
| Qwen3-4B | Q8_0 | 4.3 GB | 97.5 estimate |
| gemma-4-31B-it | UD-IQ2_XXS | 8.5 GB | 48.9 estimate |
| gemma-4-12B-it-qat-q4_0-unquantized | UD-Q4_K_XL | 6.7 GB | 62.1 estimate |
| gemma-3-1b-it | Q8_0 | 1.1 GB | 348 estimate |
| gemma-4-E2B-it | Q8_0 | 5.0 GB | 84.0 estimate |
| DeepSeek-R1-0528-Qwen3-8B | Q8_0 | 8.7 GB | 47.9 estimate |
| Llama-3.2-3B-Instruct | Q8_0 | 3.4 GB | 122 estimate |
| Phi-4-mini-instruct | Q8_0 | 4.1 GB | 102 estimate |
Similar hardware
NVIDIA GTX 1080 Ti · NVIDIA RTX 3080 · NVIDIA RTX 5070 · NVIDIA TITAN V · NVIDIA TITAN Xp · NVIDIA RTX 3060 · NVIDIA RTX 4070 Ti · NVIDIA RTX 4070 SUPER