Leaderboard
Multi-dimensional rankings based on model speed tests and provider health checks. Compare providers, endpoints, and reliability at a glance.
Average time to first token. Lower is better for responsiveness.
| Rank | Provider | Model | First Token Latency | Avg tokens per second | Total Tests |
|---|---|---|---|---|---|
| 1 | Qwen/Qwen2-7B-Instruct | 0.54 s Best: 0.44Worst: 0.73 | 98.75t/s | 5 | |
| 2 | deepseek/deepseek-v3-0324 | 0.55 s Best: 0.53Worst: 0.58 | 33.69t/s | 5 |
| 3 | S Studioapi.studio.nebius.ai | deepseek-ai/DeepSeek-V3 | 0.57 s Best: 0.52Worst: 0.61 | 19.48t/s | 5 |
| 4 | x.aiapi.x.ai | grok-2-1212 | 0.60 s Best: 0.45Worst: 1.07 | 58.57t/s | 10 |
| 5 | SiliconFlowapi.siliconflow.cn | Qwen/Qwen2.5-14B-Instruct | 0.61 s Best: 0.46Worst: 1.09 | 64.55t/s | 5 |
| 6 | S Studioapi.studio.nebius.ai | deepseek-ai/DeepSeek-V3-0324-fast | 0.62 s Best: 0.57Worst: 0.78 | 73.57t/s | 5 |
| 7 | SiliconFlowapi.siliconflow.cn | Qwen/QwQ-32B-Preview | 0.62 s Best: 0.58Worst: 0.72 | 69.75t/s | 5 |
| 8 | SiliconFlowapi.siliconflow.cn | Qwen/Qwen2.5-Coder-7B-Instruct | 0.65 s Best: 0.45Worst: 0.85 | 38.30t/s | 5 |
| 9 | D Done Hubai.bestip.one | DeepSeek-R1 | 0.67 s Best: 0.64Worst: 0.69 | 37.13t/s | 5 |
| 10 | S Studioapi.studio.nebius.ai | deepseek-ai/DeepSeek-V3-0324 | 0.70 s Best: 0.53Worst: 1.14 | 10.55t/s | 5 |
| 11 | A AI Toolsplatform.aitools.cfd | zhipu/glm-4-flash | 0.72 s Best: 0.39Worst: 1.86 | 35.84t/s | 95 |
| 12 | 算 算了么 APIapi.suanli.cn | best-free | 0.74 s Best: 0.29Worst: 1.98 | 100.39t/s | 5 |
| 13 | 羊羊羊的APIllmapi.koast.top | Qwen/QwQ-32B-Preview | 0.80 s Best: 0.62Worst: 0.97 | 70.89t/s | 5 |
| 14 | 我的旅行日志openapi.llcloud.vip | deepseek-chat | 0.81 s Best: 0.67Worst: 0.96 | 23.95t/s | 5 |
| 15 | 箴 箴理科技api.truth-ai.com.cn | qwen-max-2024-01-07 | 0.82 s Best: 0.77Worst: 0.92 | 17.48t/s | 5 |
| 16 | D Done Hubai.bestip.one | qwen-qwq | 0.89 s Best: 0.66Worst: 1.41 | 26.75t/s | 5 |
| 17 | DashScopedashscope.aliyuncs.com | qwen1.5-7b-chat | 0.89 s Best: 0.64Worst: 1.84 | 59.96t/s | 10 |
| 18 | a arkark.cn-beijing.volces.com | ep-20250304104316-qpkbf | 0.90 s Best: 0.76Worst: 1.12 | 25.49t/s | 5 |
| 19 | 箴 箴理科技api.truth-ai.com.cn | qwen-turbo-2024-11-01 | 0.92 s Best: 0.79Worst: 0.99 | 50.95t/s | 5 |
| 20 | 速 速创APIapi.mengzuihuawu.xyz | deepseek-chat | 0.93 s Best: 0.68Worst: 1.42 | 37.51t/s | 5 |
| 21 | A AI Toolsplatform.aitools.cfd | zhipu/glm-4v-flash | 0.95 s Best: 0.78Worst: 1.27 | 77.34t/s | 5 |
| 22 | F FineOneAPIit-ai.fineres.com:3000 | deepseek-v3-0324 | 0.96 s Best: 0.90Worst: 1.18 | 36.85t/s | 5 |
| 23 | F FineOneAPIit-ai.fineres.com:3000 | deepseek-v3-0324 | 0.96 s Best: 0.90Worst: 1.18 | 36.85t/s | 5 |
| 24 | DashScopedashscope.aliyuncs.com | qwen-plus | 0.97 s Best: 0.64Worst: 1.95 | 21.50t/s | 5 |
| 25 | BytesBoostapi.bytesboost.com | gpt-4o | 1.03 s Best: 0.68Worst: 1.39 | 78.41t/s | 5 |
| 26 | MN APIwww.mnapi.com | QwQ-32B | 1.09 s Best: 0.82Worst: 1.73 | 297.86t/s | 5 |
| 27 | b binaryYukiai.tzpro.xyz | gpt-4o | 1.11 s Best: 0.84Worst: 1.55 | 114.07t/s | 10 |
| 28 | Koyebunemployed-loreen-smnet-145256-d256ebc0.koyeb.app | deepseek-chat | 1.12 s Best: 0.78Worst: 1.39 | 20.87t/s | 5 |
| 29 | SiliconFlowapi.siliconflow.cn | Qwen/Qwen2.5-Coder-32B-Instruct | 1.18 s Best: 0.63Worst: 2.83 | 23.58t/s | 5 |
| 30 | New APIapi.lianwusuoai.top | 火山V3 | 1.26 s Best: 0.79Worst: 2.50 | 38.75t/s | 5 |
| 31 | N Newagiaiapi.newagiai.com | gpt-4o-mini | 1.27 s Best: 0.94Worst: 1.49 | 86.04t/s | 5 |
| 32 | a arkark.cn-beijing.volces.com | ep-20250213224710-j4lcg | 1.27 s Best: 0.87Worst: 1.58 | 19.99t/s | 5 |
| 33 | o oneapi.mrhua.toponeapi.mrhua.top | qwen2.5-mlx | 1.29 s Best: 0.89Worst: 2.29 | 12.58t/s | 5 |
| 34 | MN APIwww.mnapi.com | gemini-2.0-flash | 1.30 s Best: 1.14Worst: 1.44 | 220.95t/s | 5 |
| 35 | a api-q8w9c2d9u4m9cb54.aistudio-app.comapi-q8w9c2d9u4m9cb54.aistudio-app.com | qwq | 1.30 s Best: 0.61Worst: 3.25 | 26.57t/s | 5 |
| 36 | 派 派欧算力云api.ppinfra.com | deepseek/deepseek-v3-turbo | 1.31 s Best: 0.68Worst: 2.90 | 28.27t/s | 15 |
| 37 | 百 百万APIapi.zhubaiwan.xyz | Groq/qwen-qwq-32b | 1.36 s Best: 1.05Worst: 2.41 | 382.11t/s | 5 |
| 38 | Hugging Facehuggingface.co | qwen-max-latest | 1.40 s Best: 1.19Worst: 1.94 | 32.90t/s | 5 |
| 39 | SiliconFlowapi.siliconflow.cn | deepseek-ai/DeepSeek-V3 | 1.40 s Best: 0.70Worst: 3.88 | 17.57t/s | 20 |
| 40 | ChatAnywhereapi.chatanywhere.tech | gpt-3.5-turbo | 1.41 s Best: 1.25Worst: 1.81 | 121.17t/s | 5 |
| 41 | MN APIwww.mnapi.com | deepseek-v3-0324 | 1.47 s Best: 1.01Worst: 1.70 | 33.21t/s | 5 |
| 42 | YUNWU APIyunwu.ai | deepseek-v3-0324 | 1.49 s Best: 1.27Worst: 1.78 | 24.84t/s | 5 |
| 43 | KKSJ-AIapi.kksj.org | gpt-4o-2024-11-20 | 1.52 s Best: 1.15Worst: 1.90 | 88.57t/s | 5 |
| 44 | 箴 箴理科技api.truth-ai.com.cn | jinshu-1.5 | 1.59 s Best: 1.00Worst: 3.68 | 66.48t/s | 10 |
| 45 | KKSJ-AIapi.kksj.org | grok-3 | 1.59 s Best: 1.43Worst: 1.92 | 40.75t/s | 5 |
| 46 | 箴 箴理科技api.truth-ai.com.cn | jinshu | 1.62 s Best: 1.10Worst: 3.26 | 56.96t/s | 5 |
| 47 | N Newagiaiapi.newagiai.com | deepseek-v3-0324 | 1.64 s Best: 1.16Worst: 1.91 | 20.36t/s | 5 |
| 48 | ChatAnywhereapi.chatanywhere.tech | gpt-4o-mini | 1.64 s Best: 1.40Worst: 1.93 | 86.37t/s | 5 |
| 49 | 8 8.217.45.96:30028.217.45.96:3002 | gpt-4o | 1.71 s Best: 0.85Worst: 3.92 | 92.54t/s | 10 |
| 50 | KKSJ-AIapi.kksj.org | gemini-2.0-pro-exp-02-05 | 1.77 s Best: 1.51Worst: 2.19 | 49.44t/s | 5 |