GPUs tracked29Price history3 dayssince Sep 3, 2026Price points (24h)10,787 pointsBiggest 24h drop—Biggest 24h rise—Data updated Sep 5, 2026

Popular Open-Weight Models for Local Deployment

Model family

Updated 4 hours ago
VersionVendorLicense
Qwen3 8BAlibabaApr 27, 20258.28.27GeForce RTX 4080— · 16 GB13,232,997Apache-2.0
Qwen3.5 9BAlibabaFeb 27, 20269.79.78GeForce RTX 4080— · 16 GB12,375,273Apache-2.0
Gemma 4 31BGoogle DeepMindMar 11, 202631.331.326Arc Pro B65— · 32 GB8,332,852Apache-2.0
Gemma 4 26B A4BGoogle DeepMindMar 11, 202625.846plus 13 GB of RAMRadeon RX 7900 XT— · 20 GB8,211,474Apache-2.0
gpt-oss 20BOpenAIAug 4, 202520.93.64plus 12 GB of RAMGeForce RTX 4080— · 16 GB6,448,506Apache-2.0
Qwen3.8 27BAlibabaAug 5, 202627.827.819Radeon RX 7900 XT— · 20 GB5,739,341Apache-2.0
Llama 3.1 8BMeta AIJul 18, 2024887GeForce RTX 4080— · 16 GB5,734,979Llama 3.1 Community License
Qwen3.6 27BAlibabaApr 21, 202627.827.819Radeon RX 7900 XT— · 20 GB5,366,737Apache-2.0
gpt-oss 120BOpenAIAug 4, 2025116.85.14plus 65 GB of RAMRTX PRO 5000 Blackwell 72GB— · 72 GB5,259,501Apache-2.0
Qwen3 32BAlibabaApr 27, 202532.832.822Arc Pro B60— · 24 GB5,024,271Apache-2.0
Gemma 4 E4BGoogle DeepMindMar 2, 2026887GeForce RTX 4080— · 16 GB4,850,749Apache-2.0
DeepSeek V4 FlashDeepSeekJul 31, 2026284137plus 155 GB of RAMMac Studio M3 Ultra 256GB— · 256 GB4,559,659MIT
Qwen3.6 35B A3BAlibabaApr 15, 20263634plus 19 GB of RAMArc Pro B60— · 24 GB4,546,612Apache-2.0
Gemma 4 12BGoogle DeepMindMay 23, 2026121211GeForce RTX 4080— · 16 GB3,195,490Apache-2.0
Mistral 7B v0.3Mistral AIMay 22, 20247.37.37GeForce RTX 4080— · 16 GB2,648,636Apache-2.0
Kimi K3Moonshot AIJun 13, 20262,80010433plus 1,531 GB of RAMA100 40GB PCIe— · 40 GB · Experts offloaded2,639,566Kimi K3 License
GLM 4.7 FlashZ.ai / Zhipu AIJan 19, 20263034plus 17 GB of RAMRadeon RX 7900 XT— · 20 GB1,935,018MIT
DeepSeek V3.2DeepSeekDec 1, 20256713712plus 365 GB of RAMMac Studio M3 Ultra 512GB— · 512 GB1,559,697MIT
Llama 3.2 3BMeta AISep 18, 20243.23.24GeForce RTX 4080— · 16 GB1,419,885Llama 3.2 Community License
GLM 5.2Z.ai / Zhipu AIJun 16, 20267444041plus 406 GB of RAMMac Studio M3 Ultra 512GB— · 512 GB1,134,389MIT
DeepSeek R1 Qwen3 8BDeepSeekMay 29, 20258.28.27GeForce RTX 4080— · 16 GB1,003,991MIT
Llama 3.3 70BMeta AINov 26, 202470.670.643GeForce RTX 4090 48GB (modded)— · 48 GB815,592Llama 3.3 Community License
GLM 5.3 FlashZ.ai / Zhipu AIAug 25, 20263201830plus 174 GB of RAMMac Studio M3 Ultra 256GB— · 256 GB654,957MIT
Kimi K2.6Moonshot AIApr 14, 20261,000329plus 552 GB of RAMGeForce RTX 4080— · 16 GB · Experts offloaded638,370Modified MIT
DeepSeek R1 Distill Qwen 32BDeepSeekJan 20, 202532.832.822Arc Pro B60— · 24 GB562,818MIT
Kimi K2.5Moonshot AIJan 1, 20261,000329plus 552 GB of RAMGeForce RTX 4080— · 16 GB · Experts offloaded518,180Modified MIT
DeepSeek R1 Distill Qwen 14BDeepSeekJan 20, 202514.814.811GeForce RTX 4080— · 16 GB396,404MIT
Mistral NemoMistral AIJul 17, 202412.312.310GeForce RTX 4080— · 16 GB358,821Apache-2.0
GLM 5.3Z.ai / Zhipu AIAug 25, 20267444041plus 406 GB of RAMMac Studio M3 Ultra 512GB— · 512 GB303,534GLM-5.3 License
Ministral 3 14BMistral AIOct 31, 2025141411GeForce RTX 4080— · 16 GB266,145Apache-2.0
Kimi Linear 48B A3BMoonshot AIOct 30, 202549.133plus 27 GB of RAMArc Pro B65— · 32 GB187,057MIT
Llama 4 Scout 17B 16EMeta AIApr 2, 2025108.61710plus 55 GB of RAMInstinct MI210— · 64 GB174,923Llama 4 Community License
Ministral 3 8BMistral AIOct 31, 20258.98.98GeForce RTX 4080— · 16 GB142,610Apache-2.0
Mistral Small 3.2 24BMistral AIJun 19, 2025242416GeForce RTX 4080— · 16 GB126,611Apache-2.0
GLM 4.5 AirZ.ai / Zhipu AIJul 20, 2025106127plus 56 GB of RAMInstinct MI210— · 64 GB114,264MIT
gpt-oss Safeguard 20BOpenAISep 18, 202520.93.64plus 12 GB of RAMGeForce RTX 4080— · 16 GB78,266Apache-2.0
Kimi K2Moonshot AISep 3, 20251,000329plus 552 GB of RAMGeForce RTX 4080— · 16 GB · Experts offloaded37,308Modified MIT

Minimum VRAM = Q4_K_M weights + 8K context + runtime overhead · Used prices are median asking prices, not sold prices · Methodology