Best LLMs for 16 GB of VRAM
| Text models: 247 | Visual models: 18 | Sized for RTX 5060 Ti 16 GB | Strongest first |
Text models for 16 GB
| Model | Best fit | Context | Speed | Strongest at |
|---|---|---|---|---|
| Qwen3.8 27B | 3-bit | 20K | 23 tok/s | Chat, Coding, Agents, Docs |
| Muse Glimmer 30B | 3-bit | 40K | 22 tok/s | Docs |
| Gemma 4 26B A4B | 3-bit | 224K | 129 tok/s | Chat, Coding, Agents, Docs |
| Qwen3.5 9B | 8-bit | 144K | 33 tok/s | Chat, Coding, Agents, Docs |
| Gemma 4 12B | 8-bit | 248K | 24 tok/s | Chat, Coding, Docs |
| Qwen3.5 4B | 8-bit | 256K | 59 tok/s | Agents, Docs |
| Devstral Small 2 24B | 3-bit | 20K | 27 tok/s | Coding |
| LFM2.5 24B A2B | 4-bit | 44K | 160 tok/s | Agents |
| Mistral Small 3.2 24B | 3-bit | 20K | 27 tok/s | Chat, Coding, Agents |
| Ministral 3 14B | 6-bit | 20K | 28 tok/s | Coding, Agents, Docs |
| Gemma 4 E4B | 8-bit | 128K | 53 tok/s | Chat, Coding, Docs |
| gpt-oss 20B | MXFP4 | 120K | 104 tok/s | Agents |
| Ministral 3 8B | 8-bit | 40K | 34 tok/s | Docs |
| Gemma 4 E2B | 8-bit | 128K | 96 tok/s | Chat |
| Nemotron 3 Nano 4B | 8-bit | 256K | 74 tok/s | Docs |
| Llama 3.1 8B | 8-bit | 48K | 38 tok/s | Chat, Coding, Agents, Docs |
| Qwen3 0.6B | 8-bit | 40K | 295 tok/s | Not rated |
| Qwen3 14B | 6-bit | 16K | 26 tok/s | Not rated |
| Qwen3 VL 8B Instruct | 8-bit | 40K | 35 tok/s | Text and images |
| Qwen3 8B | 8-bit | 40K | 37 tok/s | Not rated |
| Qwen2.5 7B Instruct | 8-bit | 32K | 40 tok/s | Not rated |
| Qwen2.5 VL 7B Instruct | 8-bit | 112K | 37 tok/s | Text and images |
| Unlimited OCR | 8-bit | 32K | 218 tok/s | Reading documents |
| Qwen3 VL 4B Instruct | 8-bit | 64K | 62 tok/s | Text and images |
| Qwen3 1.7B | 8-bit | 40K | 129 tok/s | Not rated |
| Qwen3 4B | 8-bit | 40K | 68 tok/s | Not rated |
| Qwen3 4B Instruct 2507 | 8-bit | 72K | 68 tok/s | Not rated |
| Qwen2.5 3B Instruct | 8-bit | 32K | 87 tok/s | Not rated |
| Qwen3.6 27B | 3-bit | 20K | 23 tok/s | Text and images |
| DeepSeek OCR | 8-bit | 8K | 218 tok/s | Reading documents |
| Qwen2.5 Coder 14B Instruct | 6-bit | 16K | 26 tok/s | Coding |
| Qwen2.5 VL 3B Instruct | 8-bit | 125K | 72 tok/s | Text and images |
| Qwen3.5 27B | 3-bit | 20K | 23 tok/s | Text and images |
| Qwen2.5 Coder 7B Instruct | 8-bit | 32K | 40 tok/s | Coding |
| Mistral 7B Instruct V0.2 | 8-bit | 32K | 40 tok/s | Not rated |
| Chandra OCR 2 | 8-bit | 256K | 52 tok/s | Reading documents |
| Qwen3 VL 2B Instruct | 8-bit | 112K | 124 tok/s | Text and images |
| MiniCPM5 2B | 8-bit | 128K | 113 tok/s | Not rated |
| Gemma 3 4B IT | 8-bit | 128K | 75 tok/s | Text and images |
| Llama 3.2 3B Instruct | 8-bit | 96K | 85 tok/s | Not rated |
| Qwen2.5 14B Instruct | 6-bit | 16K | 26 tok/s | Not rated |
| Qwen3 1.7B Base | 8-bit | 32K | 150 tok/s | Not rated |
| Diffusiongemma 26B A4B IT | 3-bit | 112K | 124 tok/s | Text and images |
| HunyuanOCR | 8-bit | 128K | 222 tok/s | Text and images |
| Qwen3 0.6B Base | 8-bit | 32K | 350 tok/s | Not rated |
| Qwen2 VL 7B Instruct | 8-bit | 32K | 37 tok/s | Text and images |
| Ornith 1.5 9B | 8-bit | 144K | 33 tok/s | Not rated |
| Unlimited OCR 2 | 8-bit | 128K | 217 tok/s | Reading documents |
| Granite 4.1 3B | 8-bit | 128K | 79 tok/s | Not rated |
| Mistral 7B V0.1 | 8-bit | 32K | 40 tok/s | Not rated |
| Qwen2.5 Coder 3B Instruct | 8-bit | 32K | 87 tok/s | Coding |
| Meta Llama 3 8B Instruct Abliterated V3 | 8-bit | 8K | 38 tok/s | Not rated |
| EXAONE 3.5 7.8B Instruct | 8-bit | 32K | 38 tok/s | Not rated |
| Saiga Llama3 8B | 8-bit | 8K | 38 tok/s | Not rated |
| DeepSeek R1 0528 Qwen3 8B | 8-bit | 44K | 37 tok/s | Not rated |
| Gemma 3 12B IT | 8-bit | 28K | 25 tok/s | Text and images |
| Phi 3.5 Mini Instruct | 8-bit | 28K | 74 tok/s | Not rated |
| Qwen3 4B Thinking 2507 | 8-bit | 72K | 68 tok/s | Not rated |
| Phi 4 Mini Instruct | 8-bit | 80K | 72 tok/s | Not rated |
| Dots.ocr | 8-bit | 128K | 97 tok/s | Reading documents |
Fits, context and speed on the RTX 5060 Ti 16 GB. A slower card with the same memory holds the same models at lower speed. Ranked by our use-case scores, then by Hugging Face downloads. The top 60 of 247 are shown.
Image and video models
| Model | Best fit | Per output | Strongest at |
|---|---|---|---|
| LTX-2.5 | 5-bit | 2.6 min | Video |
| Z-Image Turbo | 16-bit | 20 s | Images |
| FLUX.1 [dev] | 8-bit | 39 s | Images |
| Qwen-Image | 5-bit | 14 min | Images |
| Qwen-Image 2.1 | 16-bit | 5.2 min | Images |
| FLUX.3 Klein | 16-bit | 29 s | Images |
| Wan2.2 I2V A14B | 4-bit | 2.1 h | Video |
| FLUX.2 [klein] 9B | 8-bit | 9.7 s | Images |
| Wan2.1 I2V 14B | 4-bit | 2.1 h | Video |
| Wan 2.1 T2V-14B | 4-bit | 2.6 h | Video |
| Krea 2 Turbo | 8-bit | 38 s | Images |
| Stable Diffusion 3.5 Large | 8-bit | 2.7 min | Images |
| Qwen Image 2512 | 5-bit | 14 min | Images |
| MOSS-Image | 16-bit | 3.4 min | Images |
| Wan 2.2 TI2V-5B | 16-bit | 29 min | Video |
| MOSS Image Base | 16-bit | 3.4 min | Images |
| Wan 2.2 A14B | 4-bit | 2.1 h | Video |
| HunyuanVideo 1.5 | 8-bit | 2.2 h | Video |
Machines with 16 GB
Questions
What is the best LLM for 16 GB of VRAM?
Of the open models we track, the strongest that run well on 16 GB are Qwen3.8 27B, Muse Glimmer 30B, Gemma 4 26B A4B. 247 text models fit in all.
Calculated from published specs, not benchmarked. Every machine and tier