RunsOnMy ModelsGPUs & MacsAPI pricesVRAM calculatorNew

What AI models run on the RTX 4080 Super?

The RTX 4080 Super has 16 GB of VRAM (736 GB/s). About 15.2 GB is usable for a model. 194 of 285 open models we track fit at 8K context.

194models fit (8K ctx)
180still fit at 32K ctx
40Blargest that fits
736 GB/smemory bandwidth

Models that fit, largest first

ModelParamsBest quant (8K)Needs32K contextMax tok/s (est.)
flint40B (1.3B act.)Q2_K14.8 GBno1570details
Maxwell V1 Flash37B (2.3B act.)Q2_K14.0 GBQ2_K864details
Qwen3.6 35B A3B36B (3.0B act.)Q2_K13.6 GBQ2_K654details
Ornith 1.5 35B A3B36B (3.0B act.)Q2_K13.2 GBQ2_K674details
Qwen3.5 35B A3B36B (3.0B act.)Q2_K13.6 GBQ2_K654details
Axion 1.0 35B35B (3.9B act.)Q2_K13.3 GBQ2_K503details
Qwen AgentWorld 35B A3B35B (3.0B act.)Q2_K13.1 GBQ2_K654details
Yi 1.5 34B Chat34BQ2_K14.3 GBno57details
Nous Hermes 2 Yi 34B34BQ2_K14.8 GBno55details
DeepSeek R1 Distill Qwen 32B33BQ2_K14.2 GBno60details
Qwen3 32B33BQ2_K14.1 GBno60details
GLM Z1 32B 041433BQ2_K12.6 GBQ2_K60details
GLM 4 32B 041433BQ2_K12.6 GBQ2_K60details
granite 4.0 h small32B (8.8B act.)Q2_K11.7 GBQ2_K229details
gemma 4 31B it31BQ2_K13.9 GBno63details
GLM 4.7 Flash31B (4.6B act.)Q2_K11.7 GBQ2_K440details
Qwen3 Coder 30B A3B Instruct31B (3.0B act.)Q2_K11.9 GBQ2_K665details
North Mini Code 1.030B (3.3B act.)Q2_K12.0 GBQ2_K594details
Hy Memory 30B A3B30B (3.0B act.)Q2_K11.9 GBQ2_K654details
Muse Glimmer 30B30BQ2_K11.4 GBQ2_K66details
granite 4.2 30b29BQ2_K12.7 GBno68details
Qwen3.8 27B28BQ2_K11.1 GBQ2_K71details
Qwen3.6 27B28BQ2_K11.1 GBQ2_K71details
Qwen3.5 27B28BQ2_K11.1 GBQ2_K71details
gemma 3 27b it27BQ2_K11.7 GBQ2_K70details
Arete 27B27BQ3_K_M13.6 GBQ2_K56details
MiniMax H3 Image PE27BQ2_K11.3 GBQ2_K69details
Trinity Mini26B (2.0B act.)Q3_K_M12.3 GBQ3_K_M808details
GLiDE 27B26BQ3_K_M13.2 GBQ2_K58details
diffusiongemma 26B A4B it26B (4.0B act.)Q3_K_M13.2 GBQ2_K376details
gemma 4 26B A4B it26B (4.0B act.)Q3_K_M13.3 GBQ2_K373details
LFM2.5 24B A2B24B (2.0B act.)Q3_K_M11.7 GBQ3_K_M753details
ERNIE 4.5 21B A3B PT22BQ4_K_M13.5 GBQ3_K_M55details
ERNIE 4.5 21B A3B Thinking22BQ4_K_M13.5 GBQ3_K_M55details
gpt oss safeguard 20b22B (4.8B act.)Q8_012.3 GBQ8_0273details
gpt oss 20b21B (4.2B act.)Q8_012.3 GBQ8_0303details
Groot 19B A3B 26102319B (3.0B act.)Q4_K_M12.7 GBQ4_K_M382details
Kimi VL A3B Thinking 250616B (3.0B act.)Q6_K13.5 GBQ5_K_M299details
Kimi VL A3B Instruct16B (3.0B act.)Q5_K_M12.2 GBQ5_K_M334details
Kimi VL A3B Thinking16B (3.0B act.)Q6_K13.5 GBQ5_K_M299details
deepseek moe 16b base16B (2.8B act.)Q4_K_M12.5 GBQ2_K393details
LLaDA2.1 mini16B (1.4B act.)Q6_K13.4 GBQ5_K_M626details
Moonlight 16B A3B Instruct16B (3.0B act.)Q5_K_M12.2 GBQ5_K_M325details
DeepSeek Coder V2 Lite Instruct16B (2.7B act.)Q5_K_M11.9 GBQ5_K_M367details
DeepSeek V2 Lite16B (2.7B act.)Q5_K_M11.3 GBQ5_K_M388details
DeepSeek V2 Lite Chat16B (2.7B act.)Q5_K_M11.9 GBQ5_K_M367details
Qwen2.5 Coder 14B Instruct15BQ6_K13.5 GBQ3_K_M61details
DeepSeek R1 Distill Qwen 14B15BQ6_K13.5 GBQ3_K_M61details
Qwen3 14B15BQ6_K13.2 GBQ3_K_M61details
phi 415BQ6_K13.4 GBQ3_K_M61details
gemma 3 12b it12BQ8_013.4 GBQ6_K59details
Llama Guard 4 12B12BQ6_K11.5 GBQ4_K_M75details
Llama 3.2 11B Vision Instruct11BQ8_011.6 GBQ5_K_M71details
Llama 3.2 11B Vision11BQ8_012.4 GBQ5_K_M65details
GLM 4.6V Flash10BQ8_010.3 GBQ8_074details
GLM 4.1V 9B Thinking10BQ8_010.3 GBQ8_074details
Step3 VL 10B10BQ8_09.9 GBQ8_085details
SenseNova Loop 9B Preview9.7BQ8_010.0 GBQ8_077details
Qwen3.5 9B9.7BQ8_010.0 GBQ8_077details
Ornith 1.5 9B9.7BQ8_010.0 GBQ8_077details
Qwen3.5 9B Base9.7BQ8_010.0 GBQ8_077details
MiMo V2.6 Distill Qwen 9B9.4BQ8_010.0 GBQ8_077details
glm 4 9b9.4BQ6_K13.4 GBno89details
internlm3 8b instruct8.8BQ8_09.7 GBQ8_079details
granite 4.2 8b8.8BQ8_010.5 GBQ6_K79details
granite 4.1 8b8.8BQ8_010.5 GBQ6_K79details
Qwen3 VL 8B Instruct8.8BQ8_09.9 GBQ8_085details
Molmo2 8B8.7BQ8_09.9 GBQ8_085details
LFM2.5 8B A1B8.5B (1.0B act.)Q8_09.1 GBQ8_0692details
Qwen2.5 VL 7B Instruct8.3BQ8_08.7 GBQ8_091details
UI TARS 1.5 7B8.3BQ8_08.7 GBQ8_091details
olmOCR 2 7B 10258.3BQ8_08.7 GBQ8_091details
olmOCR 7B 0225 preview8.3BQ8_08.7 GBQ8_091details
UI TARS 7B DPO8.3BQ8_08.7 GBQ8_091details
Qwen3 8B8.2BQ8_09.9 GBQ8_085details
DeepSeek R1 0528 Qwen3 8B8.2BQ8_09.9 GBQ8_085details
granite 3.3 8b instruct8.2BQ8_09.8 GBQ8_085details
granite 3.1 8b instruct8.2BQ8_09.8 GBQ8_085details
granite 3.2 8b instruct8.2BQ8_09.8 GBQ8_085details
Apertus 8B Instruct 25098.1BQ8_09.6 GBQ8_086details
Llama 3.1 8B Instruct8.0BQ8_09.6 GBQ8_086details
Llama 3.1 8B8.0BQ8_09.6 GBQ8_086details
Hermes 3 Llama 3.1 8B8.0BQ8_09.6 GBQ8_086details
DeepSeek R1 Distill Llama 8B8.0BQ8_09.6 GBQ8_086details
Meta Llama 3 8B Instruct8.0BQ8_09.6 GBQ8_086details
Meta Llama Guard 2 8B8.0BQ8_09.6 GBQ8_086details
Meta Llama 3.1 8B Instruct8.0BQ8_09.6 GBQ8_086details
Ling 3.0 tiny7.9B (1.4B act.)Q8_08.7 GBQ8_0500details
MiMo 7B RL7.8BQ8_09.4 GBQ8_091details
internlm2 5 7b chat7.7BQ8_09.2 GBQ8_089details
internlm2 chat 7b7.7BQ8_09.2 GBQ8_089details
Qwen2.5 7B Instruct7.6BQ8_08.7 GBQ8_091details
Qwen2.5 Coder 7B Instruct7.6BQ8_08.7 GBQ8_091details
DeepSeek R1 Distill Qwen 7B7.6BQ8_08.7 GBQ8_091details
Falcon H1R 7B7.6BQ8_08.5 GBQ8_091details
Olmo 3 7B Instruct7.3BQ8_010.3 GBQ8_095details
Olmo 3 1025 7B7.3BQ8_010.3 GBQ8_095details
Olmo 3 7B Think7.3BQ8_010.3 GBQ8_095details
Mistral 7B v0.17.2BQ8_08.6 GBQ8_096details
Mistral 7B Instruct v0.27.2BQ8_08.6 GBQ8_096details
Mistral 7B Instruct v0.17.2BQ8_08.6 GBQ8_096details
falcon 7b7.2BQ8_012.1 GBno96details
falcon 7b instruct7.2BQ8_011.8 GBno100details
granite 4.0 h tiny6.9B (1.5B act.)F1613.6 GBQ8_0251details
OLMoE 1B 7B 09246.9B (1.3B act.)Q8_08.4 GBQ8_0540details
deepseek coder 7b instruct v1.56.9BQ8_011.2 GBno100details
deepseek coder 6.7b instruct6.7BQ8_011.1 GBno103details
deepseek coder 6.7b base6.7BQ8_011.1 GBno103details
Llama 2 7b hf6.7BQ8_011.1 GBno103details
Yi 6B 200K6.1BF1612.3 GBQ8_061details
Yi 6B6.1BF1612.3 GBQ8_061details
Molmo2 4B4.9BF1610.0 GBF1683details
WeVisOne VL 4B4.8BF1610.0 GBF1683details
Mage VL Cowork4.7BF1610.0 GBF1683details
Qianfan OCR4.7BF1610.1 GBF1683details
Qwen3.5 4B4.7BF169.0 GBF1687details
S5 nano4.7BF169.2 GBF1685details
LightOnOCR 3 4B4.5BF169.0 GBF1687details
SingVL S1 4B4.5BF169.0 GBF1687details
Intern Decision 4B4.5BF169.0 GBF1687details
Qwen3 VL 4B Instruct4.4BF169.3 GBF1691details
gemma 3 4b it4.3BF168.4 GBF1695details
medgemma 1.5 4b it4.3BF168.4 GBF1695details
medgemma 4b it4.3BF168.4 GBF1695details
gemma 3 4b pt4.3BF169.2 GBF1686details
Phi 3.5 vision instruct4.1BF1611.2 GBQ4_K_M89details
Qwen3 4B4.0BF169.3 GBF1691details
Qwen3 4B Instruct 25074.0BF169.3 GBF1691details
granite vision 4.1 4b4.0BF167.6 GBF16108details
Phi 4 mini instruct3.8BF168.9 GBF1696details
Phi 3 mini 4k instruct3.8BF1610.6 GBQ5_K_M96details
Phi 3.5 mini instruct3.8BF1610.6 GBQ5_K_M96details
Phi 3 mini 128k instruct3.8BF1610.6 GBQ5_K_M96details
Qwen2.5 VL 3B Instruct3.8BF166.7 GBF16119details
granite 4.2 3b3.7BF168.0 GBF16100details
granite 4.1 3b llmon preview3.7BF168.0 GBF16101details
Raon 1 nano3.5BF167.0 GBF16117details
granite 4.1 3b3.4BF167.6 GBF16108details
granite 4.0 micro3.4BF167.6 GBF16108details
DeepSeek OCR 23.4B (1.2B act.)F166.6 GBF16355details
deepseek vl2 tiny3.4B (1.2B act.)F167.4 GBF16313details
Unlimited OCR 23.3B (1.1B act.)F166.6 GBF16364details
Unlimited OCR3.3B (1.1B act.)F166.6 GBF16366details
DeepSeek OCR3.3B (1.1B act.)F166.6 GBF16366details
granite 3.1 3b a800m instruct3.3B (883M act.)F167.1 GBF16416details
Llama 3.2 3B Instruct3.2BF1613.5 GBQ8_057details
Llama 3.2 3B3.2BF167.5 GBF16115details
Hermes 3 Llama 3.2 3B3.2BF167.5 GBF16114details
d1 3B3.1BF1610.8 GBF1668details
LFM2.5 VL 3B3.1BF1610.8 GBF1668details
Qwen2.5 3B Instruct3.1BF166.7 GBF16119details
SmolLM3 3B3.1BF166.9 GBF16120details
SmolLM3 3B Base3.1BF166.9 GBF16120details
phi 22.8BF168.2 GBQ8_0132details
LFM2.5 2.6B2.7BF1610.8 GBF1668details
granite 3.3 2b instruct2.5BF165.8 GBF16145details
MiniCPM5 2B2.5BF165.7 GBF16146details
WeVisOne VL 2B2.4BF165.4 GBF16181details
Qwen3.5 2B2.3BF164.5 GBF16195details
SmolVLM Instruct2.2BF165.4 GBF16203details
Qwen3 1.7B2.0BF164.8 GBF16213details
Qwen3.8 27B DSpine b162.0BF164.8 GBF16181details
d1 2B1.9BF167.2 GBF16106details
LFM3 VL 2B1.9BF167.2 GBF16106details
DeepSeek R1 Distill Qwen 1.5B1.8BF164.2 GBF16207details
LFM3 1.7B1.7BF167.2 GBF16106details
LFM3 1.7B Base1.7BF164.0 GBF16213details
SmolLM2 1.7B Instruct1.7BF165.2 GBF16215details
SmolLM2 1.7B1.7BF165.2 GBF16215details
SmolLM 1.7B1.7BF165.2 GBF16215details
granite 4.0 1b1.6BF164.3 GBF16225details
Qwen2.5 1.5B Instruct1.5BF163.8 GBF16238details
OLMo 2 0425 1B1.5BF164.4 GBF16248details
deepseek coder 1.3b instruct1.3BF164.5 GBF16273details
Llama 3.2 1B1.2BF163.2 GBF16298details
Llama 3.2 1B Instruct1.2BF165.5 GBF16148details
LFM2.5 1.2B Instruct1.2BF165.0 GBF16157details
HunyuanOCR1.1BF163.5 GBF16329details
PaddleOCR VL 1.6959MF162.5 GBF16384details
Qwen3.5 0.8B873MF162.4 GBF16485details
LightOnOCR 3 0.8B853MF162.4 GBF16485details
FastVLM 0.5B759MF162.2 GBF16485details
Qwen3 0.6B752MF162.7 GBF16614details
GOT OCR2 0716MF162.8 GBF16514details
Nemotron 3.5 Parse631MF161.9 GBF16563details
GOT OCR 2.0 hf561MF162.5 GBF16657details
SmolVLM2 500M Video Instruct507MF161.6 GBF16897details
SmolVLM 500M Instruct507MF161.6 GBF16897details
Qwen2.5 0.5B Instruct494MF161.7 GBF16745details
LFM2.5 VL 450M449MF161.9 GBF16517details
SmolLM2 360M Instruct362MF161.5 GBF161014details
SmolLM2 360M362MF161.5 GBF161014details
LFM2.5 350M354MF161.9 GBF16517details
LFM2 350M354MF161.4 GBF161038details

Just too big at 4-bit

These need up to 35% more memory than the RTX 4080 Super 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
Qwen3.8 27B16.8 GB1.6 GB
Qwen3.5 27B17.0 GB1.8 GB
Qwen3.6 27B17.0 GB1.8 GB
diffusiongemma 26B A4B it17.1 GB1.9 GB
gemma 4 26B A4B it17.2 GB2.0 GB
gemma 3 27b it17.3 GB2.1 GB
Arete 27B17.4 GB2.2 GB
Muse Glimmer 30B17.9 GB2.7 GB
GLM 4.7 Flash18.2 GB3.0 GB
Hy Memory 30B A3B18.5 GB3.3 GB
Qwen3 Coder 30B A3B Instruct18.7 GB3.5 GB
granite 4.0 h small18.9 GB3.7 GB

Questions

What is the biggest model that runs on the RTX 4080 Super?

Of the models we track, the largest that fits at 8K context is flint (40B) at Q2_K, needing about 14.8 GB.

How much of the RTX 4080 Super'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.