Providers
119
clusters
2966
machines
10830
GPUs
111
independent providers
| cluster | hardware | model | ttft | single | throughput↓ | cache hit |
|---|---|---|---|---|---|---|
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 0.38s | 142 tps | 611 tps | 96% |
| 131× | 8x NVIDIA GeForce RTX 5090 +1 configs | qwen3.6-35b-a3b | 0.47s | 112 tps | 206 tps | 85% |
| 168× | ?x GPU +2 configs | qwen3.6-35b-a3b | 0.66s | 146 tps | 200 tps | 90% |
| 2× | 8x RTX 5090 (32GB) | deepseek-v4-flash-0731 | 1.00s | 46 tps | 154 tps | — |
| 56× | 8x NVIDIA H200 +2 configs | qwen3.6-35b-a3b | 0.90s | 156 tps | 140 tps | 94% |
| 81× | 8x NVIDIA L20 +1 configs | qwen3.6-35b-a3b | 0.70s | 126 tps | 137 tps | 84% |
| 30× | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition +1 configs | qwen3.6-35b-a3b | 1.00s | 83 tps | 134 tps | 89% |
| 283× | 8x NVIDIA RTX 6000 Ada Generation +6 configs | qwen3.6-35b-a3b | 0.68s | 57 tps | 130 tps | 87% |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | 1.12s | 152 tps | 120 tps | 86% |
| 16× | 8x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 0.83s | 181 tps | 108 tps | 78% |
| 48× | 8x NVIDIA GeForce RTX 4090 +1 configs | qwen3.6-35b-a3b | 1.82s | 115 tps | 92 tps | 83% |
| 64× | 8x NVIDIA GeForce RTX 4090 | qwen3.6-35b-a3b | 1.56s | 116 tps | 88 tps | 89% |
| 34× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 0.69s | 97 tps | 68 tps | 96% |
| 34× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 2.59s | 87 tps | 56 tps | 3% |
| 25× | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition | qwen3.6-35b-a3b | 1.40s | 144 tps | 54 tps | 98% |
| 28× | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition +1 configs | qwen3.6-35b-a3b | 1.06s | 91 tps | 52 tps | 91% |
| 1× | 8x RTX 5090 (32GB) | deepseek-v4-flash-0731 | 1.17s | 52 tps | 52 tps | — |
| 13× | 1x NVIDIA RTX PRO 6000 Blackwell Server Edition +4 configs | qwen3.6-35b-a3b | 0.58s | 121 tps | 50 tps | 90% |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | 1.92s | 143 tps | 39 tps | 100% |
| 64× | 8x NVIDIA GeForce RTX 4090 +1 configs | qwen3.6-35b-a3b | 1.27s | 128 tps | 39 tps | — |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 1.51s | 173 tps | 36 tps | 98% |
| 1× | 80x RTX 5090 (32GB) | kimi-k3 | 2.18s | 28 tps | 30 tps | 98% |
| 64× | 8x NVIDIA GeForce RTX 4090 +1 configs | qwen3.6-35b-a3b | 0.80s | 144 tps | 28 tps | — |
| 32× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | 27 tps | 4% |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | 27 tps | 70% |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | 27 tps | 7% |
| 25× | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition | qwen3.6-35b-a3b | — | — | 27 tps | — |
| 64× | ?x GPU | qwen3.6-35b-a3b | 0.63s | 166 tps | 20 tps | 29% |
| 40× | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition +1 configs | qwen3.6-35b-a3b | 21.26s | 98 tps | 20 tps | 57% |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | 20 tps | 4% |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | 20 tps | 47% |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | 2.75s | 70 tps | 18 tps | 64% |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | 16 tps | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | 14 tps | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | 14 tps | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | 0.95s | 195 tps | 10 tps | 95% |
| 16× | 8x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 0.70s | 139 tps | 8 tps | 56% |
| 1× | 80x RTX 5090 (32GB) | kimi-k3 | — | — | 7 tps | 66% |
| 21× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 0.98s | 183 tps | 7 tps | 84% |
| 32× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 2.38s | 69 tps | 6 tps | 65% |
| 1× | 32x RTX 5090 (32GB) | glm-5.2 | 3.27s | 49 tps | 6 tps | 61% |
| 36× | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition +1 configs | qwen3.6-35b-a3b | 1.11s | 222 tps | 5 tps | 54% |
| 1× | 32x RTX 5090 (32GB) | glm-5.2 | 1.36s | 61 tps | 2 tps | 73% |
| 32× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 1.27s | 96 tps | 2 tps | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | 0.40s | 238 tps | 1 tps | — |
| 31× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | 0.35s | 56 tps | 1 tps | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | 0.87s | 102 tps | 0 tps | — |
| 1× | 32x RTX 5090 (32GB) | glm-5.2 | — | — | 0 tps | 13% |
| 1× | 32x RTX 5090 (32GB) | glm-5.2 | 4.32s | — | 0 tps | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | 0 tps | — |
| 10× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 5× | 8x NVIDIA GeForce RTX 5090 | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 8x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 32× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 21× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 34× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 34× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 23× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 33× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 1× | 4x NVIDIA GeForce RTX 4090 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 2× | 1x NVIDIA H100 80GB HBM3 +1 configs | qwen3.6-35b-a3b | — | — | — | — |
| 31× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 25× | 8x NVIDIA RTX PRO 6000 Blackwell Server Edition | qwen3.6-35b-a3b | — | — | — | — |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 2× | 1x NVIDIA RTX 6000 Ada Generation | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 32× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 4x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 44× | 2x NVIDIA H100 80GB HBM3 +1 configs | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA B300 SXM6 AC | qwen3.6-35b-a3b | — | — | — | — |
| 36× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 8x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 8x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 8x RTX 5090 | qwen3.6-35b-a3b | — | — | — | — |
| 21× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 22× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 32× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 13× | 8x NVIDIA GeForce RTX 5090 +1 configs | qwen3.6-35b-a3b | — | — | — | — |
| 21× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 33× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 1× | 4x NVIDIA GeForce RTX 4090 | qwen3.6-35b-a3b | — | — | — | — |
| 1× | 8x NVIDIA GeForce RTX 4090 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 5× | 8x NVIDIA GeForce RTX 5090 | qwen3.6-35b-a3b | — | — | — | — |
| 18× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 1× | 1x NVIDIA RTX 6000 Ada Generation | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 4x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 4x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 34× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 33× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 33× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 20× | 1x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 27× | 8x NVIDIA H200 | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 4x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 8× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 33× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 1× | 1x NVIDIA GeForce RTX 4090 | qwen3.6-35b-a3b | — | — | — | — |
| 32× | 2x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 1× | 8x NVIDIA H100 80GB HBM3 | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |
| 2× | 1x NVIDIA L40S | qwen3.6-35b-a3b | — | — | — | — |
| 16× | 1x NVIDIA RTX PRO 6000 Blackwell Workstation Edition | qwen3.6-35b-a3b | — | — | — | — |