RunsOnMy ModelsGPUs & MacsBy VRAMBest GPUCan I run…?OllamaStart hereAPI pricesVRAM calculatorNew

What AI models run on the RTX 5070 Ti?

The RTX 5070 Ti has 16 GB of VRAM (896 GB/s). About 15.2 GB is usable for a model. 200 of 298 open models we track fit at 8K context.

200models fit (8K ctx)
189still fit at 32K ctx
37Blargest that fits
896 GB/smemory bandwidth

This card has 16 GB of VRAM. See every model that fits in 16 GB VRAM. Other 16 GB hardware: RTX 4060 Ti 16GB, RTX 4070 Ti Super, RTX 4080, RTX 4080 Super, RTX 5060 Ti 16GB, RTX 5080, RX 7800 XT, RX 7900 GRE.

Compare for local LLMs: RTX 5070 Ti vs RTX 5080RTX 5070 Ti vs RTX 4070 Ti SuperRTX 5070 Ti vs RX 9070 XT all comparisons

Using a specific tool? Ollama models that run on the RTX 5070 Ti · LM Studio · llama.cpp offload

Models that fit, largest first

ModelParamsBest quant (8K)Needs32K contextMax tok/s (est.)Details
Maxwell V1 Flash37B (2.3B act.)Q2_K14.0 GBQ2_K1052details
Qwen3.6 35B A3B36B (3.0B act.)Q2_K13.6 GBQ2_K795details
Ornith 1.5 35B A3B36B (3.0B act.)Q2_K13.2 GBQ2_K821details
Qwen3.5 35B A3B36B (3.0B act.)Q2_K13.6 GBQ2_K795details
Sage A1.536B (4.7B act.)Q2_K13.4 GBQ2_K512details
Qwen AgentWorld 35B A3B35B (3.0B act.)Q2_K13.1 GBQ2_K796details
Yi 1.5 34B Chat34BQ2_K14.3 GBno70details
Nous Hermes 2 Yi 34B34BQ2_K14.8 GBno67details
DeepSeek R1 Distill Qwen 32B33BQ2_K14.2 GBno73details
Qwen3 32B33BQ2_K14.1 GBno73details
GLM Z1 32B 041433BQ2_K12.5 GBQ2_K73details
GLM 4 32B 041433BQ2_K12.6 GBQ2_K73details
Olmo 3 32B Think32BQ2_K13.0 GBQ2_K75details
Granite 4.0 H Small32B (8.8B act.)Q2_K11.7 GBQ2_K278details
Gemma 4 31B IT31BQ2_K14.7 GBno71details
Gemma 4 31B31BQ2_K13.9 GBno76details
GLM 4.7 Flash31B (4.6B act.)Q2_K11.4 GBQ2_K550details
Qwen3 Coder 30B A3B Instruct31B (3.0B act.)Q2_K12.1 GBQ2_K796details
North Mini Code 1.030B (3.3B act.)Q2_K11.7 GBQ2_K745details
Hy Memory 30B A3B30B (3.0B act.)Q2_K11.9 GBQ2_K796details
Muse Glimmer 30B30BQ2_K11.4 GBQ2_K80details
Granite 4.2 30B29BQ2_K12.7 GBno83details
Raon 1 Mini Preview29B (4.3B act.)Q2_K11.2 GBQ2_K553details
Qwen3.8 27B28BQ2_K11.5 GBQ2_K83details
Qwen3.6 27B28BQ2_K12.6 GBQ2_K74details
Qwen3.5 27B28BQ2_K12.6 GBQ2_K74details
Gemma 3 27B IT27BQ2_K11.7 GBQ2_K85details
Arete 27B27BQ3_K_M13.6 GBQ2_K68details
MiniMax H3 Image PE27BQ2_K11.3 GBQ2_K84details
Trinity Mini26B (2.0B act.)Q3_K_M12.3 GBQ3_K_M984details
GLiDE 27B26BQ3_K_M13.2 GBQ2_K70details
DiffusionGemma 26B A4B IT26B (4.0B act.)Q3_K_M13.2 GBQ2_K458details
Gemma 4 26B A4B IT26B (4.0B act.)Q3_K_M13.5 GBQ2_K444details
LFM2.5 24B A2B24B (2.0B act.)Q3_K_M11.7 GBQ3_K_M917details
ERNIE 4.5 21B A3B PT22BQ4_K_M13.6 GBQ3_K_M66details
ERNIE 4.5 21B A3B Thinking22BQ4_K_M13.6 GBQ3_K_M66details
gpt-oss Safeguard 20B22B (4.8B act.)Q8_012.3 GBQ8_0332details
gpt-oss 20B21B (4.2B act.)Q8_012.3 GBQ8_0369details
Kimi VL A3B Thinking 250616B (3.0B act.)Q6_K13.5 GBQ5_K_M364details
Kimi VL A3B Thinking16B (3.0B act.)Q6_K13.5 GBQ5_K_M364details
Kimi VL A3B Instruct16B (3.0B act.)Q5_K_M12.2 GBQ5_K_M407details
DeepSeek MoE 16B Base16B (2.8B act.)Q4_K_M12.5 GBQ2_K478details
LLaDA2.1 Mini16B (1.4B act.)Q6_K13.4 GBQ5_K_M763details
Moonlight 16B A3B Instruct16B (3.0B act.)Q5_K_M12.2 GBQ5_K_M396details
DeepSeek Coder V2 Lite Instruct16B (2.7B act.)Q5_K_M11.9 GBQ5_K_M446details
DeepSeek V2 Lite16B (2.7B act.)Q6_K12.8 GBQ6_K411details
DeepSeek V2 Lite Chat16B (2.7B act.)Q5_K_M11.9 GBQ5_K_M446details
Qwen2.5 Coder 14B Instruct15BQ6_K13.5 GBQ3_K_M74details
DeepSeek R1 Distill Qwen 14B15BQ6_K13.5 GBQ3_K_M74details
Qwen3 14B15BQ6_K13.2 GBQ3_K_M74details
Phi 415BQ6_K13.4 GBQ2_K74details
Gemma 3 12B IT12BQ8_013.4 GBQ6_K72details
Llama Guard 4 12B12BQ6_K11.5 GBQ4_K_M91details
KORMo Web 11B11BQ8_013.0 GBQ5_K_M75details
Llama 3.2 11B Vision Instruct11BQ8_011.6 GBQ5_K_M86details
Llama 3.2 11B Vision11BQ8_012.4 GBQ5_K_M79details
GLM 4.6V Flash10BQ8_010.3 GBQ8_090details
GLM 4.1V 9B Thinking10BQ8_010.3 GBQ8_090details
Step3 VL 10B10BQ8_09.9 GBQ8_0103details
SenseNova Loop 9B Preview9.7BQ8_010.0 GBQ8_094details
Qwen3.5 9B9.7BQ8_010.3 GBQ8_091details
Ornith 1.5 9B9.7BQ8_010.0 GBQ8_094details
SenseNova Loop 9B9.4BQ8_010.0 GBQ8_094details
MiMo V2.6 Distill Qwen 9B9.4BQ8_010.0 GBQ8_094details
GLM 4 9B9.4BQ8_010.3 GBQ8_090details
Apertus v1.5 8B8.9BQ8_010.5 GBQ8_095details
InternLM3 8B Instruct8.8BQ8_09.7 GBQ8_096details
Granite 4.2 8B8.8BQ8_010.5 GBQ6_K96details
Granite 4.1 8B8.8BQ8_010.5 GBQ6_K96details
Granite 4.1 8B Llmon Preview8.8BQ8_010.5 GBQ6_K96details
Qwen3 VL 8B Instruct8.8BQ8_09.9 GBQ8_0103details
Molmo2 8B8.7BQ8_09.9 GBQ8_0103details
LFM2.5 8B A1B8.5B (1.0B act.)Q8_09.1 GBQ8_0842details
Qwen2.5 VL 7B Instruct8.3BQ8_08.7 GBQ8_0111details
UI TARS 1.5 7B8.3BQ8_08.7 GBQ8_0111details
olmOCR 2 7B 10258.3BQ8_08.7 GBQ8_0111details
olmOCR 7B 0225 Preview8.3BQ8_08.7 GBQ8_0111details
UI TARS 7B DPO8.3BQ8_08.7 GBQ8_0111details
Qwen3 8B8.2BQ8_09.9 GBQ8_0103details
DeepSeek R1 0528 Qwen3 8B8.2BQ8_09.9 GBQ8_0103details
Granite 3.3 8B Instruct8.2BQ8_09.8 GBQ8_0103details
Granite 3.1 8B Instruct8.2BQ8_09.8 GBQ8_0103details
Granite 3.2 8B Instruct8.2BQ8_09.8 GBQ8_0103details
Apertus 8B Instruct 25098.1BQ8_09.6 GBQ8_0105details
Llama 3.1 8B Instruct8.0BQ8_09.6 GBQ8_0105details
Llama 3.1 8B8.0BQ8_09.6 GBQ8_0105details
Hermes 3 Llama 3.1 8B8.0BQ8_09.6 GBQ8_0105details
Llama Guard 3 8B8.0BQ8_09.6 GBQ8_0105details
DeepSeek R1 Distill Llama 8B8.0BQ8_09.6 GBQ8_0105details
Meta Llama 3 8B Instruct8.0BQ8_09.6 GBQ8_0105details
Meta Llama Guard 2 8B8.0BQ8_09.6 GBQ8_0105details
Meta Llama 3.1 8B Instruct8.0BQ8_09.6 GBQ8_0105details
Aya Expanse 8B8.0BQ8_09.8 GBQ8_0105details
Ling 3.0 Tiny7.9B (1.4B act.)Q8_08.7 GBQ8_0609details
MiMo 7B RL7.8BQ8_09.4 GBQ8_0111details
InternLM2.5 7B Chat7.7BQ8_09.2 GBQ8_0109details
InternLM2 Chat 7B7.7BQ8_09.2 GBQ8_0109details
Qwen2.5 7B Instruct7.6BQ8_08.7 GBQ8_0111details
DeepSeek R1 Distill Qwen 7B7.6BQ8_08.7 GBQ8_0111details
Qwen2.5 Coder 7B Instruct7.6BQ8_08.7 GBQ8_0111details
Olmo 3 7B Instruct7.3BQ8_010.3 GBQ8_0115details
Olmo 3 1025 7B7.3BQ8_010.3 GBQ8_0115details
Olmo 3 7B Think7.3BQ8_010.3 GBQ8_0115details
Mistral 7B v0.17.2BQ8_08.6 GBQ8_0116details
Mistral 7B Instruct v0.27.2BQ8_08.6 GBQ8_0116details
Mistral 7B Instruct v0.17.2BQ8_08.6 GBQ8_0116details
Falcon 7B7.2BQ8_07.7 GBQ8_0117details
Falcon 7B Instruct7.2BQ8_07.4 GBQ8_0122details
Granite 4.0 H Tiny6.9B (1.5B act.)F1613.6 GBQ8_0305details
OLMoE 1B 7B 09246.9B (1.3B act.)Q8_08.4 GBQ8_0657details
DeepSeek Coder 7B Instruct v1.56.9BQ8_011.2 GBno122details
DeepSeek Coder 6.7B Instruct6.7BQ8_011.1 GBno125details
DeepSeek Coder 6.7B Base6.7BQ8_011.1 GBno125details
Llama 2 7B HF6.7BQ8_011.1 GBno125details
Yi 6B 200K6.1BF1612.3 GBQ8_074details
Yi 6B6.1BF1612.3 GBQ8_074details
Molmo2 4B4.9BF1610.0 GBF16101details
WeVisOne VL 4B4.8BF1610.0 GBF16101details
Mage VL Cowork4.7BF1610.0 GBF16101details
Qianfan OCR4.7BF1610.1 GBF16101details
Qwen3.5 4B4.7BF169.2 GBF16103details
SingVL S1 4B4.5BF169.0 GBF16106details
LightOnOCR 3 4B4.5BF169.0 GBF16106details
Intern Decision 4B4.5BF169.0 GBF16106details
Qwen3 VL 4B Instruct4.4BF169.3 GBF16111details
Gemma 3 4B IT4.3BF168.4 GBF16115details
MedGemma 1.5 4B IT4.3BF168.4 GBF16115details
MedGemma 4B IT4.3BF168.4 GBF16115details
TranslateGemma 4B IT4.3BF169.2 GBF16104details
Gemma 3 4B PT4.3BF168.4 GBF16115details
Phi 3.5 Vision Instruct4.1BF1611.2 GBQ4_K_M108details
Qwen3 4B4.0BF169.3 GBF16111details
Qwen3 4B Instruct 25074.0BF169.3 GBF16111details
Granite Vision 4.1 4B4.0BF167.6 GBF16132details
Phi 4 Mini Instruct3.8BF168.9 GBF16117details
Phi 3 Mini 4K Instruct3.8BF1610.6 GBQ5_K_M117details
Phi 3.5 Mini Instruct3.8BF1610.6 GBQ5_K_M117details
Phi 3 Mini 128K Instruct3.8BF1610.6 GBQ5_K_M117details
Qwen2.5 VL 3B Instruct3.8BF168.0 GBF16119details
Granite 4.2 3B3.7BF168.0 GBF16122details
Granite 4.1 3B Llmon Preview3.7BF168.0 GBF16122details
Raon 1 Nano3.5BF167.0 GBF16143details
Granite 4.1 3B3.4BF167.6 GBF16132details
Granite 4.0 Micro3.4BF167.6 GBF16132details
DeepSeek OCR 23.4B (1.2B act.)F166.6 GBF16433details
DeepSeek VL2 Tiny3.4B (1.2B act.)F167.4 GBF16381details
Tiny Aya Global3.3BF167.5 GBF16134details
Unlimited OCR 23.3B (1.1B act.)F166.6 GBF16443details
Unlimited OCR3.3B (1.1B act.)F166.6 GBF16446details
DeepSeek OCR3.3B (1.1B act.)F166.6 GBF16446details
Granite 3.1 3B A800M Instruct3.3B (883M act.)F167.1 GBF16507details
Llama 3.2 3B Instruct3.2BF167.5 GBF16139details
Llama 3.2 3B3.2BF167.5 GBF16139details
Hermes 3 Llama 3.2 3B3.2BF167.5 GBF16139details
d1 3B3.1BF165.8 GBF16166details
LFM2.5 VL 3B3.1BF165.8 GBF16166details
Qwen2.5 3B Instruct3.1BF166.7 GBF16145details
SmolLM3 3B3.1BF166.9 GBF16145details
SmolLM3 3B Base3.1BF166.9 GBF16146details
Phi 22.8BF168.2 GBQ8_0161details
LFM2.5 2.6B2.7BF1610.8 GBF1683details
Granite 3.3 2B Instruct2.5BF165.8 GBF16177details
MiniCPM5 2B2.5BF165.7 GBF16178details
WeVisOne VL 2B2.4BF165.4 GBF16220details
Cosmos Reason2 2B2.4BF165.4 GBF16220details
Qwen3.5 2B2.3BF164.6 GBF16230details
SmolVLM Instruct2.2BF165.4 GBF16247details
Qwen3 1.7B2.0BF165.4 GBF16220details
d1 2B1.9BF164.0 GBF16258details
LFM3 VL 2B1.9BF164.0 GBF16258details
DeepSeek R1 Distill Qwen 1.5B1.8BF164.2 GBF16252details
LFM3 1.7B1.7BF167.2 GBF16129details
LFM3 1.7B Base1.7BF164.0 GBF16258details
SmolLM2 1.7B Instruct1.7BF165.2 GBF16262details
SmolLM2 1.7B1.7BF165.2 GBF16262details
SmolLM 1.7B1.7BF165.2 GBF16262details
Granite 4.0 1B1.6BF164.3 GBF16274details
Qwen2.5 1.5B Instruct1.5BF163.8 GBF16290details
OLMo 2 0425 1B1.5BF164.4 GBF16301details
DeepSeek Coder 1.3B Instruct1.3BF164.5 GBF16333details
Llama 3.2 1B1.2BF163.2 GBF16361details
Llama 3.2 1B Instruct1.2BF163.2 GBF16361details
LFM2.5 1.2B Instruct1.2BF165.0 GBF16191details
LFM2 1.2B1.2BF163.0 GBF16382details
HunyuanOCR1.1BF163.5 GBF16400details
PaddleOCR VL 1.6959MF162.5 GBF16467details
Qwen3.5 0.8B873MF162.4 GBF16575details
LightOnOCR 3 0.8B853MF162.4 GBF16591details
FastVLM 0.5B759MF162.2 GBF16590details
Qwen3 0.6B752MF163.0 GBF16594details
GOT OCR2.0716MF162.8 GBF16626details
Nemotron 3.5 Parse631MF161.9 GBF16686details
GOT OCR 2.0 HF561MF162.5 GBF16799details
SmolVLM2 500M Video Instruct507MF161.6 GBF161092details
SmolVLM 500M Instruct507MF161.6 GBF161092details
Qwen2.5 0.5B Instruct494MF161.7 GBF16907details
LFM2.5 VL 450M449MF161.3 GBF161259details
SmolLM2 360M Instruct362MF161.5 GBF161235details
SmolLM2 360M362MF161.5 GBF161235details
LFM2.5 350M354MF161.9 GBF16630details

Bigger models with CPU offload

These popular models don't fit in the RTX 5070 Ti's VRAM at 4-bit, but run with llama.cpp offload, keeping part of the model in system RAM. Speed range is DDR4-3200 to DDR5-6000 (estimate). Method.

ModelParamsOffload plan
GLM 5.3 Flash321B MoEOffload Q2_K, 128 GB RAM · ~7.5–15 tok/s
gpt-oss 120B117B MoEOffload Q4_K_M, 64 GB RAM · ~14–29 tok/s
DeepSeek V4 Flash 0731304B MoEOffload Q2_K, 128 GB RAM · ~8.4–16 tok/s
DeepSeek V4 Flash291B MoEOffload Q3_K_M, 128 GB RAM · ~7.3–14 tok/s
GLM 4.5 Air110B MoEOffload Q4_K_M, 64 GB RAM · ~5.8–11 tok/s
MiniMax M2.5229B MoEOffload Q4_K_M, 128 GB RAM · ~5.7–11 tok/s
Hy3299B MoEOffload Q3_K_M, 128 GB RAM · ~4.2–8.5 tok/s

Comparing cards? The RTX 5070 Ti (launch MSRP $749) is in best GPUs for local LLMs under $800. See also Mac vs PC.

Just too big at 4-bit

These need up to 35% more memory than the RTX 5070 Ti has at Q4_K_M and 8K context. A smaller quant or shorter context may work.

ModelQ4_K_M needsShort by
Trinity Mini15.8 GB0.6 GB
GLiDE 27B16.2 GB1.0 GB
MiniMax H3 Image PE16.8 GB1.6 GB
DiffusionGemma 26B A4B IT17.1 GB1.9 GB
Gemma 4 26B A4B IT17.3 GB2.1 GB
Gemma 3 27B IT17.3 GB2.1 GB
Arete 27B17.4 GB2.2 GB
Qwen3.8 27B17.6 GB2.4 GB
Muse Glimmer 30B17.9 GB2.7 GB
Qwen3.5 27B18.1 GB2.9 GB
Qwen3.6 27B18.1 GB2.9 GB
GLM 4.7 Flash18.3 GB3.1 GB
Hy Memory 30B A3B18.5 GB3.3 GB
Qwen3 Coder 30B A3B Instruct18.8 GB3.6 GB
North Mini Code 1.018.8 GB3.6 GB

Questions

What is the biggest model that runs on the RTX 5070 Ti?

Of the models we track, the largest that fits at 8K context is Maxwell V1 Flash (37B) at Q2_K, needing about 14.0 GB.

How much of the RTX 5070 Ti's memory can a model use?

About 15.2 GB: we assume 95% of the 16 GB is usable after driver and display overhead.

Spec source: NVIDIA specifications. Model data: Hugging Face Hub API. Speed is a bandwidth ceiling (bandwidth ÷ bytes read per token), not a benchmark. Methodology · Report an error.