OpenWeightsTerminal
Earn

Payback

Calendar time until the street price comes back. US residential power. No decode, no date.

US prices, Sep 2026. Power is 18.34¢/kWh, EIA June 2026. ROI/yr is at 20% duty. Color is a log scale. An asterisk marks an outlier.

Street stacks

  • 2× Dual

    DeepSeek v4.1 Flash · 103 tok/s · $9,998

    5% 109 yr · 70% 7.8 yr · ROI/yr 3.7%

    27 yr20% duty

  • 4× Quad

    GLM 5.3 Flash · 104 tok/s · $19,996

    5% 370 yr · 70% 26 yr · ROI/yr 1.1%

    92 yr20% duty

  • 1× Solo

    DGX Spark · $4,999

    5% — · 70% — · ROI/yr —

    No decode20% duty

  • 3× Triple

    DGX Spark · $14,997

    5% — · 70% — · ROI/yr —

    No decode20% duty

  • 6× 6+

    DGX Spark · $29,994

    5% — · 70% — · ROI/yr —

    No decode20% duty

Quoted pairs

  • Qwen3.8-27B

    RTX 5090 · 370 tok/s · $4,639

    5% 11 yr · 70% 9.5 mo · ROI/yr 36%

    2.8 yr20% duty

  • Ternary Bonsai 2 27B

    RTX 5060 Ti 16GB · 54.7 tok/s · $614

    5% 11 yr · 70% 9.6 mo · ROI/yr 36%

    2.8 yr20% duty

  • Qwen3.5 2B

    RTX 5090 32GB · 351 tok/s · $4,639

    5% 12 yr · 70% 10 mo · ROI/yr 34%

    2.9 yr20% duty

  • Qwen3.8-27B

    RTX 5090 · 258 tok/s · $4,639

    5% 17 yr · 70% 14 mo · ROI/yr 24%

    4.2 yr20% duty

  • Swift 1.5 Qwen3.8 Flash-Next

    RTX 4090 24GB · 220 tok/s · $4,495

    5% 19 yr · 70% 16 mo · ROI/yr 21%

    4.7 yr20% duty

  • Qwen3.6-35B-A3B

    M4 Pro · 85.5 tok/s · $2,199

    5% 21 yr · 70% 18 mo · ROI/yr 19%

    5.3 yr20% duty

  • Qwen3.5-4B

    M4 Pro · 82.8 tok/s · $2,199

    5% 22 yr · 70% 19 mo · ROI/yr 18%

    5.5 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 5090 · 195.8 tok/s · $4,639

    5% 23 yr · 70% 20 mo · ROI/yr 17%

    5.8 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 5090 · 191 tok/s · $4,639

    5% 24 yr · 70% 20 mo · ROI/yr 17%

    6.0 yr20% duty

  • Qwen3.8-27B

    RTX 5090 · 186 tok/s · $4,639

    5% 25 yr · 70% 21 mo · ROI/yr 16%

    6.2 yr20% duty

  • Gemma-4 26B-A4B-it

    RTX 5090 32GB · 173 tok/s · $4,639

    5% 27 yr · 70% 23 mo · ROI/yr 15%

    6.7 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 5090 32GB · 142 tok/s · $4,639

    5% 35 yr · 70% 2.5 yr · ROI/yr 11%

    8.7 yr20% duty

  • Qwen3.8-Flash-Next

    M5 Max · 126.5 tok/s · $5,399

    5% 35 yr · 70% 2.5 yr · ROI/yr 11%

    8.8 yr20% duty

  • Qwen3.5-9B

    M4 Pro · 49.3 tok/s · $2,199

    5% 38 yr · 70% 2.7 yr · ROI/yr 11%

    9.4 yr20% duty

  • Qwen3-8B

    M4 Pro · 48.3 tok/s · $2,199

    5% 38 yr · 70% 2.7 yr · ROI/yr 10%

    9.6 yr20% duty

  • Qwen3.8-27B

    RTX 5060 Ti 16GB · 23 tok/s · $614

    5% 42 yr · 70% 3.0 yr · ROI/yr 9.5%

    11 yr20% duty

  • Qwen3.8-27B

    DGX Spark · 102.9 tok/s · $4,999

    5% 45 yr · 70% 3.2 yr · ROI/yr 8.8%

    11 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 5090 · 114.3 tok/s · $4,639

    5% 47 yr · 70% 3.4 yr · ROI/yr 8.4%

    12 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 5090 32GB · 104 tok/s · $4,639

    5% 55 yr · 70% 3.9 yr · ROI/yr 7.3%

    14 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 4090 · 92 tok/s · $4,495

    5% 56 yr · 70% 4.0 yr · ROI/yr 7.1%

    14 yr20% duty

  • Nemotron 3.5 Bolt 74B

    DGX Spark · 80 tok/s · $4,999

    5% 61 yr · 70% 4.4 yr · ROI/yr 6.5%

    15 yr20% duty

  • Qwen3.8-Flash-Next

    DGX Spark · 79.5 tok/s · $4,999

    5% 62 yr · 70% 4.4 yr · ROI/yr 6.5%

    15 yr20% duty

  • Qwen3.8-27B

    RTX 5060 Ti 16GB · 19 tok/s · $614

    5% 65 yr · 70% 4.6 yr · ROI/yr 6.2%

    16 yr20% duty

  • Qwen3.8-Flash-Next

    DGX Spark · 74.8 tok/s · $4,999

    5% 67 yr · 70% 4.8 yr · ROI/yr 6.0%

    17 yr20% duty

  • Qwen3.8-27B

    DGX Spark · 72 tok/s · $4,999

    5% 70 yr · 70% 5.0 yr · ROI/yr 5.7%

    17 yr20% duty

  • AliceAI-Foundation-80B-A3B

    M5 Max 128GB · 65 tok/s · $5,399

    5% 71 yr · 70% 5.0 yr · ROI/yr 5.7%

    18 yr20% duty

  • Qwen3.8-Flash-Next

    DGX Spark · 65 tok/s · $4,999

    5% 80 yr · 70% 5.7 yr · ROI/yr 5.0%

    20 yr20% duty

  • Qwen3.8-Flash-Next

    DGX Spark · 65 tok/s · $4,999

    5% 80 yr · 70% 5.7 yr · ROI/yr 5.0%

    20 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 5090 32GB · 80.9 tok/s · $4,639

    5% 83 yr · 70% 5.9 yr · ROI/yr 4.8%

    21 yr20% duty

  • Qwen3.8-Flash-Next Coder

    M5 Max 64GB · 51 tok/s · $5,399

    5% 92 yr · 70% 6.6 yr · ROI/yr 4.3%

    23 yr20% duty

  • CYBER-FROST 3.8

    DGX Spark · 54.7 tok/s · $4,999

    5% 101 yr · 70% 7.2 yr · ROI/yr 4.0%

    25 yr20% duty

  • Underdog Saluki 27B

    DGX Spark · 53 tok/s · $4,999

    5% 105 yr · 70% 7.5 yr · ROI/yr 3.8%

    26 yr20% duty

  • DeepSeek v4.1 Flash

    2× DGX Spark · 103 tok/s · $9,998

    5% 109 yr · 70% 7.8 yr · ROI/yr 3.7%

    27 yr20% duty

  • Qwen3.8-Flash-Next

    DGX Spark · 50.2 tok/s · $4,999

    5% 114 yr · 70% 8.1 yr · ROI/yr 3.5%

    28 yr20% duty

  • Qwen3.6-35B-A3B

    2× DGX Spark · 94.4 tok/s · $9,998

    5% 124 yr · 70% 8.9 yr · ROI/yr 3.2%

    31 yr20% duty

  • DeepSeek v4.1 Flash

    2× DGX Spark · 93.6 tok/s · $9,998

    5% 126 yr · 70% 9.0 yr · ROI/yr 3.2%

    31 yr20% duty

  • GLM 5.3 Flash

    2x DGX Spark · 85 tok/s · $9,998

    5% 146 yr · 70% 10 yr · ROI/yr 2.7%

    36 yr20% duty

  • Qwen3.8-27B

    DGX Spark · 42.3 tok/s · $4,999

    5% 147 yr · 70% 10 yr · ROI/yr 2.7%

    37 yr20% duty

  • Nex-N2.5-mini

    RTX 5060 Ti 16GB · 14 tok/s · $614

    5% 192 yr · 70% 14 yr · ROI/yr 2.1%

    48 yr20% duty

  • RED-SNOW 5.3 Flash

    DGX Spark · 33.0 tok/s · $4,999

    5% 224 yr · 70% 16 yr · ROI/yr 1.8%

    56 yr20% duty

  • GLM 5.3 Flash

    DGX Spark · 32.3 tok/s · $4,999

    5% 233 yr · 70% 17 yr · ROI/yr 1.7%

    58 yr20% duty

  • Qwen3.8-Flash-Next

    DGX Spark · 32 tok/s · $4,999

    5% 237 yr · 70% 17 yr · ROI/yr 1.7%

    59 yr20% duty

  • Qwen3.8-27B

    DGX Spark · 31.8 tok/s · $4,999

    5% 240 yr · 70% 17 yr · ROI/yr 1.7%

    60 yr20% duty

  • GLM 5.3 Flash

    4× DGX Spark · 127 tok/s · $19,996

    5% 241 yr · 70% 17 yr · ROI/yr 1.7%

    60 yr20% duty

  • GLM 5.3 Flash

    2× DGX Spark · 60.4 tok/s · $9,998

    5% 266 yr · 70% 19 yr · ROI/yr 1.5%

    66 yr20% duty

  • GLM 5.3 Flash

    DGX Spark · 30 tok/s · $4,999

    5% 269 yr · 70% 19 yr · ROI/yr 1.5%

    67 yr20% duty

  • Qwen3.8-Flash-Next

    2× DGX Spark · 58 tok/s · $9,998

    5% 289 yr · 70% 21 yr · ROI/yr 1.4%

    72 yr20% duty

  • Qwen3.8-27B

    DGX Spark · 28.7 tok/s · $4,999

    5% 295 yr · 70% 21 yr · ROI/yr 1.4%

    74 yr20% duty

  • Qwen3.8-27B

    RTX 4090 24GB · 40.7 tok/s · $4,495

    5% 296 yr · 70% 21 yr · ROI/yr 1.4%

    74 yr20% duty

  • GLM 5.3 Flash

    2x DGX Spark · 55.6 tok/s · $9,998

    5% 317 yr · 70% 23 yr · ROI/yr 1.3%

    79 yr20% duty

  • GLM 5.3 Flash

    4× DGX Spark · 104 tok/s · $19,996

    5% 370 yr · 70% 26 yr · ROI/yr 1.1%

    92 yr20% duty

  • Qwen3.8-Flash-Next

    2× DGX Spark · 45 tok/s · $9,998

    5% 549 yr · 70% 39 yr · ROI/yr 0.7%

    137 yr20% duty

  • Qwen3.8-Flash-Next

    RTX 5090 · 42.1 tok/s · $4,639

    5% 671 yr · 70% 48 yr · ROI/yr 0.6%

    168 yr20% duty

  • GLM 5.3 Flash

    DGX Spark · 18.6 tok/s · $4,999

    5% 1206 yr · 70% 86 yr · ROI/yr 0.3%

    301 yr20% duty

  • Qwen3.8-Flash-Next

    M5 Max 64GB · 7.8 tok/s · $5,399

    5% 1281 yr · 70% 91 yr · ROI/yr 0.3%

    320 yr20% duty

  • Qwen3.8-27B TurboFCFusion

    RTX 5060 Ti 16GB · 10 tok/s · $614

    5% Never · 70% Never · ROI/yr -1.2%

    Outlier: Never20% duty

  • GLM 5.3

    4× DGX Spark · 35 tok/s · $19,996

    5% Never · 70% Never · ROI/yr -0.7%

    Outlier: Never20% duty

  • GLM 5.3 Flash

    2x DGX Spark · 15 tok/s · $9,998

    5% Never · 70% Never · ROI/yr -0.8%

    Outlier: Never20% duty

  • GLM 5.3

    4x DGX Spark · 30 tok/s · $19,996

    5% Never · 70% Never · ROI/yr -0.8%

    Outlier: Never20% duty

  • GLM 5.3

    4x DGX Spark · 38.4 tok/s · $19,996

    5% Never · 70% Never · ROI/yr -0.6%

    Outlier: Never20% duty

  • GLM 5.3

    8× DGX Spark · 40 tok/s · $39,992

    5% Never · 70% Never · ROI/yr -1.0%

    Outlier: Never20% duty

  • Empero Qwen3.8-35B-A3B

    RTX 3060 12GB · 50 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Ornith

    GTX 1660 SUPER 6GB · 40 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • MiMo-V2.6-Distill-Qwen-9B

    RTX 5060 8GB · 47 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    M5 Pro 64GB · 41.4 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next Coder

    RTX 5060 Ti 16GB + RTX 3070 8GB · 127 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Ternary Bonsai 2 27B

    RTX 3060 12GB · 50 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3080 Ti 12GB · 79 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3090 24GB · 70 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3090 24GB · 87 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    2× RTX 3090 24GB · 158.3 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    2× RTX 3090 24GB · 153.4 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3090 24GB · 100 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 5080 16GB · 50 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    2× AMD MI50 32GB · 52 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    AMD MI50 32GB · 26.8 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX PRO 6000 · 92.8 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • DeepSeek v4.1 Flash

    RTX PRO 6000 · 57.7 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • GLM 5.3 Flash

    4× RTX PRO 6000 · 264 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • GLM 5.3 Flash

    2× RTX PRO 6000 · 238 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.6-35B-A3B

    M4 Pro 64GB · 108.6 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-27B

    RTX PRO 6000 Blackwell Max-Q · 272.8 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-27B

    RTX PRO 6000 Blackwell Max-Q · 151.4 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3060 12GB · 27.7 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3090 24GB · 72 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.6-35B-A3B

    M1 Max 32GB · 37 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-27B

    M5 Pro 64GB · 89.9 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-27B

    M6 mini 32GB · 61.9 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-27B

    RTX A5000 · 177.6 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-27B

    RTX A5000 · 139.3 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3090 24GB · 100.6 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3080 10GB · 52.7 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 4060 Ti · 50 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • GLM 5.3 Flash

    RTX PRO 6000 + 2× GB10 · 120 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    2× RTX 5060 Ti 16GB · 99.9 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 5070 12GB · 66.7 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 5070 Ti 16GB · 53.5 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    2× RTX 3060 12GB · 40 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    Ryzen AI Max+ 395 128GB · 53.8 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    2× RTX 3090 24GB · 122.5 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • REAP50 GLM 5.3 Flash

    Radeon RX 9070 XT · 10 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • GLM 5.3

    2× DGX Station GB300 · 81.5 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    Tesla V100 32GB · 57 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • GLM 5.3 Flash

    Mac Studio M4 Max 64GB · 11.3 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    4× Tesla V100 32GB · 60 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 4060 Ti 16GB · 55 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    Intel Arc B570 10GB · 9.0 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    Intel Arc B570 10GB · 32.4 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Gemma-4 26B-A4B-it

    Radeon RX 9060 XT 16GB · 88 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 3090 + RTX 2070 SUPER · 34.2 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty

  • Qwen3.8-Flash-Next

    RTX 4060 Ti 16GB · 55 tok/s

    5% — · 70% — · ROI/yr —

    No US price20% duty