What can the RTX 3080 10 GB run?
| Text models: 244 | Visual models: 12 | Strongest first |
- Memory
- 10 GB
- Bandwidth
- 760 GB/s
- Maker
- NVIDIA
Text models that run on it
| Model | Best fit | Context | Speed | Strongest at |
|---|---|---|---|---|
| Qwen3.5 9B | 6-bit | 32K | 70 tok/s | Chat, Coding, Agents, Docs |
| Gemma 4 12B | 5-bit | 24K | 57 tok/s | Chat, Coding, Docs |
| Qwen3.5 4B | 8-bit | 120K | 96 tok/s | Agents, Docs |
| Gemma 4 E4B | 8-bit | 32K | 86 tok/s | Chat, Coding, Docs |
| Ministral 3 8B | 5-bit | 20K | 82 tok/s | Docs |
| Gemma 4 E2B | 8-bit | 128K | 152 tok/s | Chat |
| Nemotron 3 Nano 4B | 8-bit | 256K | 117 tok/s | Docs |
| Llama 3.1 8B | 6-bit | 20K | 80 tok/s | Chat, Coding, Agents, Docs |
| Qwen3 0.6B | 8-bit | 40K | 427 tok/s | Not rated |
| Qwen3 VL 8B Instruct | 5-bit | 20K | 84 tok/s | Text and images |
| Qwen3 8B | 6-bit | 16K | 78 tok/s | Not rated |
| Qwen2.5 7B Instruct | 8-bit | 20K | 67 tok/s | Not rated |
| Qwen2.5 VL 7B Instruct | 6-bit | 40K | 78 tok/s | Text and images |
| Unlimited OCR | 8-bit | 32K | 333 tok/s | Reading documents |
| Qwen2.5 1.5B Instruct | 8-bit | 32K | 255 tok/s | Not rated |
| Qwen3.5 2B | 8-bit | 256K | 189 tok/s | Text and images |
| Qwen3 VL 4B Instruct | 8-bit | 28K | 101 tok/s | Text and images |
| Qwen3 1.7B | 8-bit | 40K | 204 tok/s | Not rated |
| Qwen3 4B | 8-bit | 32K | 110 tok/s | Not rated |
| Qwen3 4B Instruct 2507 | 8-bit | 32K | 110 tok/s | Not rated |
| Qwen2.5 3B Instruct | 8-bit | 32K | 138 tok/s | Not rated |
| DeepSeek OCR | 8-bit | 8K | 333 tok/s | Reading documents |
| Qwen2.5 VL 3B Instruct | 8-bit | 125K | 117 tok/s | Text and images |
| Qwen2.5 Coder 7B Instruct | 8-bit | 20K | 67 tok/s | Coding |
| Mistral 7B Instruct V0.2 | 6-bit | 24K | 84 tok/s | Not rated |
| Chandra OCR 2 | 8-bit | 96K | 86 tok/s | Reading documents |
| Llama 3.2 1B Instruct | 8-bit | 128K | 339 tok/s | Not rated |
| Qwen3 VL 2B Instruct | 8-bit | 60K | 196 tok/s | Text and images |
| Qwen2 VL 2B Instruct | 8-bit | 32K | 190 tok/s | Text and images |
| MiniCPM5 2B | 8-bit | 128K | 174 tok/s | Not rated |
| Gemma 3 4B IT | 8-bit | 128K | 121 tok/s | Text and images |
| Llama 3.2 3B Instruct | 8-bit | 48K | 138 tok/s | Not rated |
| LightOnOCR 2 1B | 8-bit | 16K | 351 tok/s | Text and images |
| Qwen3 1.7B Base | 8-bit | 32K | 234 tok/s | Not rated |
| HunyuanOCR | 8-bit | 80K | 338 tok/s | Text and images |
| Qwen3 0.6B Base | 8-bit | 32K | 493 tok/s | Not rated |
| Rax 4.5 | 8-bit | 256K | 189 tok/s | Text and images |
| Qwen2 VL 7B Instruct | 6-bit | 32K | 78 tok/s | Text and images |
| Ornith 1.5 9B | 6-bit | 32K | 70 tok/s | Not rated |
| Unlimited OCR 2 | 8-bit | 88K | 332 tok/s | Reading documents |
| Granite 4.1 3B | 8-bit | 64K | 125 tok/s | Not rated |
| DeepSeek R1 Distill Qwen 1.5B | 8-bit | 128K | 255 tok/s | Not rated |
| Mistral 7B V0.1 | 8-bit | 32K | 66 tok/s | Not rated |
| Qwen2.5 Coder 3B Instruct | 8-bit | 32K | 138 tok/s | Coding |
| Meta Llama 3 8B Instruct Abliterated V3 | 6-bit | 8K | 80 tok/s | Not rated |
| EXAONE 3.5 7.8B Instruct | 6-bit | 20K | 81 tok/s | Not rated |
| Saiga Llama3 8B | 6-bit | 8K | 80 tok/s | Not rated |
| MiniCPM V 4.7 1B | 8-bit | 256K | 298 tok/s | Text and images |
| Qwen2.5 Coder 1.5B Instruct | 8-bit | 32K | 255 tok/s | Coding |
| DeepSeek R1 0528 Qwen3 8B | 6-bit | 16K | 78 tok/s | Not rated |
| Gemma 3 12B IT | 4-bit | 16K | 71 tok/s | Text and images |
| Phi 3.5 Mini Instruct | 5-bit | 16K | 168 tok/s | Not rated |
| Qwen3 4B Thinking 2507 | 8-bit | 32K | 110 tok/s | Not rated |
| Phi 4 Mini Instruct | 8-bit | 36K | 116 tok/s | Not rated |
| Dots.ocr | 8-bit | 128K | 155 tok/s | Reading documents |
| DeepSeek R1 Distill Qwen 7B | 8-bit | 20K | 67 tok/s | Not rated |
| Phi 3 Mini 4K Instruct | 8-bit | 4K | 119 tok/s | Not rated |
| DeepSeek OCR 2 | 8-bit | 8K | 322 tok/s | Reading documents |
| SmolLM3 3B | 8-bit | 64K | 138 tok/s | Not rated |
| Qwen2 7B Instruct | 8-bit | 20K | 67 tok/s | Not rated |
Best fit is the highest quality, up to 8-bit, that leaves the use case’s context. Ranked by our use-case scores, then by Hugging Face downloads. Speeds are rough, plus or minus 25%. The top 60 of 244 are shown.
Image and video models
| Model | Best fit | Per output | Strongest at |
|---|---|---|---|
| Stable Diffusion XL 1.0 | 16-bit | 11 s | Images |
| Z-Image Turbo | 8-bit | 16 s | Images |
| FLUX.1 [dev] | 5-bit | 36 s | Images |
| Qwen-Image 2.1 | 8-bit | 4.2 min | Images |
| FLUX.3 Klein | 8-bit | 23 s | Images |
| FLUX.2 [klein] 9B | 8-bit | 7.9 s | Images |
| Krea 2 Turbo | 5-bit | 35 s | Images |
| Stable Diffusion 3.5 Large | 8-bit | 2.2 min | Images |
| MOSS-Image | 8-bit | 2.7 min | Images |
| Wan 2.2 TI2V-5B | 8-bit | 23 min | Video |
| MOSS Image Base | 8-bit | 2.7 min | Images |
| HunyuanVideo 1.5 | 5-bit | 2.0 h | Video |
Only just too big
- LFM2.5 24B A2B
- Ministral 3 14B
- gpt-oss 20B
- Qwen3 14B
- Qwen2.5 Coder 14B Instruct
- Qwen2.5 14B Instruct
- GPT Neox 20B
- Phi 4
- DeepSeek R1 Distill Qwen 14B
- Qwen2.5 14B Instruct 1M
- CodeLlama 7B HF
- Qwen1.5 7B
- Deepseek Coder 6.7B Base
- Qwen 7B Chat
- Llama2 13B Chat HF
- Phi 4 Reasoning
- Molmo2 O 7B
- ALMA 13B R
- Llava 1.5 13B HF
- Internlm2 Chat 20B
- Glm 4 9B
- Internlm2 Chat 20B ExPO
- HyperCLOVAX SEED Think 14B
- OLMo 2 1124 13B Instruct
- Phi 3 Medium 128K Instruct
These need up to a quarter more memory than the RTX 3080 10 GB has. Spilling into system RAM or a lower precision can get them running, slowly.
Questions
What is the best LLM for the RTX 3080 10 GB?
Of the open models we track, the strongest that run well on the RTX 3080 10 GB are Qwen3.5 9B, Gemma 4 12B, Qwen3.5 4B. 244 text models fit it in all.
Calculated from published specs, not benchmarked. All machines