Skip to content

What can the RTX 5070 Ti run?

What fits on the RTX 5070 Ti
Text models: 247Visual models: 18Strongest first
Memory
16 GB
Bandwidth
896 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 5070 Ti, best first
ModelBest fitContextSpeedStrongest at
Qwen3.8 27B3-bit20K44 tok/sChat, Coding, Agents, Docs
Muse Glimmer 30B3-bit40K42 tok/sDocs
Gemma 4 26B A4B3-bit224K210 tok/sChat, Coding, Agents, Docs
Qwen3.5 9B8-bit144K64 tok/sChat, Coding, Agents, Docs
Gemma 4 12B8-bit248K46 tok/sChat, Coding, Docs
Qwen3.5 4B8-bit256K112 tok/sAgents, Docs
Devstral Small 2 24B3-bit20K52 tok/sCoding
LFM2.5 24B A2B4-bit44K231 tok/sAgents
Mistral Small 3.2 24B3-bit20K52 tok/sChat, Coding, Agents
Ministral 3 14B6-bit20K54 tok/sCoding, Agents, Docs
Gemma 4 E4B8-bit128K100 tok/sChat, Coding, Docs
gpt-oss 20BMXFP4120K181 tok/sAgents
Ministral 3 8B8-bit40K66 tok/sDocs
Gemma 4 E2B8-bit128K174 tok/sChat
Nemotron 3 Nano 4B8-bit256K135 tok/sDocs
Llama 3.1 8B8-bit48K73 tok/sChat, Coding, Agents, Docs
Qwen3 0.6B8-bit40K473 tok/sNot rated
Qwen3 14B6-bit16K51 tok/sNot rated
Qwen3 VL 8B Instruct8-bit40K67 tok/sText and images
Qwen3 8B8-bit40K72 tok/sNot rated
Qwen2.5 7B Instruct8-bit32K78 tok/sNot rated
Qwen2.5 VL 7B Instruct8-bit112K72 tok/sText and images
Unlimited OCR8-bit32K377 tok/sReading documents
Qwen3 VL 4B Instruct8-bit64K116 tok/sText and images
Qwen3 1.7B8-bit40K234 tok/sNot rated
Qwen3 4B8-bit40K127 tok/sNot rated
Qwen3 4B Instruct 25078-bit72K127 tok/sNot rated
Qwen2.5 3B Instruct8-bit32K159 tok/sNot rated
Qwen3.6 27B3-bit20K44 tok/sText and images
DeepSeek OCR8-bit8K377 tok/sReading documents
Qwen2.5 Coder 14B Instruct6-bit16K51 tok/sCoding
Qwen2.5 VL 3B Instruct8-bit125K134 tok/sText and images
Qwen3.5 27B3-bit20K44 tok/sText and images
Qwen2.5 Coder 7B Instruct8-bit32K78 tok/sCoding
Mistral 7B Instruct V0.28-bit32K77 tok/sNot rated
Chandra OCR 28-bit256K100 tok/sReading documents
Qwen3 VL 2B Instruct8-bit112K225 tok/sText and images
MiniCPM5 2B8-bit128K197 tok/sNot rated
Gemma 3 4B IT8-bit128K140 tok/sText and images
Llama 3.2 3B Instruct8-bit96K159 tok/sNot rated
Qwen2.5 14B Instruct6-bit16K51 tok/sNot rated
Qwen3 1.7B Base8-bit32K266 tok/sNot rated
Diffusiongemma 26B A4B IT3-bit112K203 tok/sText and images
HunyuanOCR8-bit128K382 tok/sText and images
Qwen3 0.6B Base8-bit32K541 tok/sNot rated
Qwen2 VL 7B Instruct8-bit32K72 tok/sText and images
Ornith 1.5 9B8-bit144K64 tok/sNot rated
Unlimited OCR 28-bit128K375 tok/sReading documents
Granite 4.1 3B8-bit128K144 tok/sNot rated
Mistral 7B V0.18-bit32K77 tok/sNot rated
Qwen2.5 Coder 3B Instruct8-bit32K159 tok/sCoding
Meta Llama 3 8B Instruct Abliterated V38-bit8K73 tok/sNot rated
EXAONE 3.5 7.8B Instruct8-bit32K74 tok/sNot rated
Saiga Llama3 8B8-bit8K73 tok/sNot rated
DeepSeek R1 0528 Qwen3 8B8-bit44K72 tok/sNot rated
Gemma 3 12B IT8-bit28K49 tok/sText and images
Phi 3.5 Mini Instruct8-bit28K138 tok/sNot rated
Qwen3 4B Thinking 25078-bit72K127 tok/sNot rated
Phi 4 Mini Instruct8-bit80K134 tok/sNot rated
Dots.ocr8-bit128K179 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 5070 Ti
ModelBest fitPer outputStrongest at
LTX-2.55-bit1.6 minVideo
Z-Image Turbo16-bit11 sImages
FLUX.1 [dev]8-bit21 sImages
Qwen-Image5-bit7.5 minImages
Qwen-Image 2.116-bit2.8 minImages
FLUX.3 Klein16-bit16 sImages
Wan2.2 I2V A14B4-bit67 minVideo
FLUX.2 [klein] 9B8-bit5.7 sImages
Wan2.1 I2V 14B4-bit67 minVideo
Wan 2.1 T2V-14B4-bit83 minVideo
Krea 2 Turbo8-bit21 sImages
Stable Diffusion 3.5 Large8-bit89 sImages
Qwen Image 25125-bit7.5 minImages
MOSS-Image16-bit1.8 minImages
Wan 2.2 TI2V-5B16-bit16 minVideo
MOSS Image Base16-bit1.8 minImages
Wan 2.2 A14B4-bit67 minVideo
HunyuanVideo 1.58-bit71 minVideo

Only just too big

Questions

What is the best LLM for the RTX 5070 Ti?

Of the open models we track, the strongest that run well on the RTX 5070 Ti 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