Skip to content

What can the RTX 5080 run?

What fits on the RTX 5080
Text models: 247Visual models: 18Strongest first
Memory
16 GB
Bandwidth
960 GB/s
Maker
NVIDIA

Open the calculator on this machineEvery model for 16 GB of VRAM

Text models that run on it

Text models that run on the RTX 5080, best first
ModelBest fitContextSpeedStrongest at
Qwen3.8 27B3-bit20K47 tok/sChat, Coding, Agents, Docs
Muse Glimmer 30B3-bit40K45 tok/sDocs
Gemma 4 26B A4B3-bit224K219 tok/sChat, Coding, Agents, Docs
Qwen3.5 9B8-bit144K68 tok/sChat, Coding, Agents, Docs
Gemma 4 12B8-bit248K49 tok/sChat, Coding, Docs
Qwen3.5 4B8-bit256K119 tok/sAgents, Docs
Devstral Small 2 24B3-bit20K55 tok/sCoding
LFM2.5 24B A2B4-bit44K238 tok/sAgents
Mistral Small 3.2 24B3-bit20K55 tok/sChat, Coding, Agents
Ministral 3 14B6-bit20K57 tok/sCoding, Agents, Docs
Gemma 4 E4B8-bit128K106 tok/sChat, Coding, Docs
gpt-oss 20BMXFP4120K190 tok/sAgents
Ministral 3 8B8-bit40K70 tok/sDocs
Gemma 4 E2B8-bit128K184 tok/sChat
Nemotron 3 Nano 4B8-bit256K143 tok/sDocs
Llama 3.1 8B8-bit48K78 tok/sChat, Coding, Agents, Docs
Qwen3 0.6B8-bit40K493 tok/sNot rated
Qwen3 14B6-bit16K55 tok/sNot rated
Qwen3 VL 8B Instruct8-bit40K72 tok/sText and images
Qwen3 8B8-bit40K77 tok/sNot rated
Qwen2.5 7B Instruct8-bit32K83 tok/sNot rated
Qwen2.5 VL 7B Instruct8-bit112K76 tok/sText and images
Unlimited OCR8-bit32K396 tok/sReading documents
Qwen3 VL 4B Instruct8-bit64K124 tok/sText and images
Qwen3 1.7B8-bit40K247 tok/sNot rated
Qwen3 4B8-bit40K134 tok/sNot rated
Qwen3 4B Instruct 25078-bit72K134 tok/sNot rated
Qwen2.5 3B Instruct8-bit32K168 tok/sNot rated
Qwen3.6 27B3-bit20K47 tok/sText and images
DeepSeek OCR8-bit8K396 tok/sReading documents
Qwen2.5 Coder 14B Instruct6-bit16K54 tok/sCoding
Qwen2.5 VL 3B Instruct8-bit125K143 tok/sText and images
Qwen3.5 27B3-bit20K47 tok/sText and images
Qwen2.5 Coder 7B Instruct8-bit32K83 tok/sCoding
Mistral 7B Instruct V0.28-bit32K82 tok/sNot rated
Chandra OCR 28-bit256K106 tok/sReading documents
Qwen3 VL 2B Instruct8-bit112K238 tok/sText and images
MiniCPM5 2B8-bit128K208 tok/sNot rated
Gemma 3 4B IT8-bit128K148 tok/sText and images
Llama 3.2 3B Instruct8-bit96K169 tok/sNot rated
Qwen2.5 14B Instruct6-bit16K54 tok/sNot rated
Qwen3 1.7B Base8-bit32K281 tok/sNot rated
Diffusiongemma 26B A4B IT3-bit112K212 tok/sText and images
HunyuanOCR8-bit128K401 tok/sText and images
Qwen3 0.6B Base8-bit32K561 tok/sNot rated
Qwen2 VL 7B Instruct8-bit32K76 tok/sText and images
Ornith 1.5 9B8-bit144K68 tok/sNot rated
Unlimited OCR 28-bit128K394 tok/sReading documents
Granite 4.1 3B8-bit128K152 tok/sNot rated
Mistral 7B V0.18-bit32K82 tok/sNot rated
Qwen2.5 Coder 3B Instruct8-bit32K168 tok/sCoding
Meta Llama 3 8B Instruct Abliterated V38-bit8K78 tok/sNot rated
EXAONE 3.5 7.8B Instruct8-bit32K79 tok/sNot rated
Saiga Llama3 8B8-bit8K78 tok/sNot rated
DeepSeek R1 0528 Qwen3 8B8-bit44K77 tok/sNot rated
Gemma 3 12B IT8-bit28K52 tok/sText and images
Phi 3.5 Mini Instruct8-bit28K146 tok/sNot rated
Qwen3 4B Thinking 25078-bit72K134 tok/sNot rated
Phi 4 Mini Instruct8-bit80K142 tok/sNot rated
Dots.ocr8-bit128K190 tok/sReading documents

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 247 are shown.

Image and video models

Image and video models that run on the RTX 5080
ModelBest fitPer outputStrongest at
LTX-2.55-bit78 sVideo
Z-Image Turbo16-bit8.9 sImages
FLUX.1 [dev]8-bit17 sImages
Qwen-Image5-bit5.9 minImages
Qwen-Image 2.116-bit2.2 minImages
FLUX.3 Klein16-bit13 sImages
Wan2.2 I2V A14B4-bit52 minVideo
FLUX.2 [klein] 9B8-bit4.6 sImages
Wan2.1 I2V 14B4-bit52 minVideo
Wan 2.1 T2V-14B4-bit65 minVideo
Krea 2 Turbo8-bit17 sImages
Stable Diffusion 3.5 Large8-bit70 sImages
Qwen Image 25125-bit5.9 minImages
MOSS-Image16-bit87 sImages
Wan 2.2 TI2V-5B16-bit13 minVideo
MOSS Image Base16-bit87 sImages
Wan 2.2 A14B4-bit52 minVideo
HunyuanVideo 1.58-bit55 minVideo

Only just too big

Questions

What is the best LLM for the RTX 5080?

Of the open models we track, the strongest that run well on the RTX 5080 are Qwen3.8 27B, Muse Glimmer 30B, Gemma 4 26B A4B. 247 text models fit it in all.

Calculated from published specs, not benchmarked. All machines