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/qwen3-32b | 0.18 s Best: 0.14Worst: 0.29 | 310.21t/s | 5 | |
| 2 | llama3.1-8b | 0.19 s Best: 0.15Worst: 0.21 | 2142.09t/s | 5 | |
| 3 |
Cerebrasapi.cerebras.ai |
| llama-3.3-70b |
0.25 s Best: 0.15Worst: 0.32 |
1532.55t/s |
| 5 |
| 4 | A AI Toolsplatform.aitools.cfd | google/gemini-2.0-flash-exp | 0.28 s Best: -Worst: 1.67 | 30.41t/s | 25 |
| 5 | Groqapi.groq.com | openai/gpt-oss-120b | 0.31 s Best: 0.25Worst: 0.36 | 456.69t/s | 10 |
| 6 | Groqapi.groq.com | openai/gpt-oss-20b | 0.47 s Best: 0.24Worst: 0.78 | 755.20t/s | 5 |
| 7 | DashScopedashscope.aliyuncs.com | qwen-flash | 0.50 s Best: 0.33Worst: 1.10 | 134.43t/s | 10 |
| 8 | Cerebrasapi.cerebras.ai | gpt-oss-120b | 0.54 s Best: 0.25Worst: 1.05 | 1920.13t/s | 5 |
| 9 | GPTAPI.USwww.gptapi.us | deepseek-v3.1 | 0.55 s Best: 0.40Worst: 0.92 | 132.30t/s | 5 |
| 10 | 智谱AI开放平台open.bigmodel.cn | GLM-4-FlashX | 0.57 s Best: 0.46Worst: 0.95 | 69.04t/s | 5 |
| 11 | a ai-hub.square-llm.comai-hub.square-llm.com | anthropic/claude-haiku-4.5 | 0.57 s Best: 0.44Worst: 0.72 | 98.63t/s | 5 |
| 12 | 全球AIglobalai.vip | gemini-2.5-flash-lite | 0.60 s Best: 0.52Worst: 0.76 | 274.78t/s | 5 |
| 13 | New APInew.123nhh.xyz | gemini-flash-lite-latest | 0.67 s Best: 0.38Worst: 0.97 | 369.22t/s | 5 |
| 14 | a api.linkapi.orgapi.linkapi.org | gemini-2.5-flash-lite-preview-06-17 | 0.70 s Best: 0.44Worst: 1.61 | 160.85t/s | 5 |
| 15 | N New APIapi.seosycy.com | deepseek-v3.2 | 0.71 s Best: 0.51Worst: 1.29 | 27.18t/s | 10 |
| 16 | free_chatgpt_apifree.v36.cm | gpt-4o-mini | 0.73 s Best: 0.55Worst: 0.96 | 116.21t/s | 10 |
| 17 | SiliconFlowapi.siliconflow.cn | tencent/Hunyuan-MT-7B | 0.74 s Best: 0.38Worst: 1.55 | 55.78t/s | 10 |
| 18 | DashScopedashscope.aliyuncs.com | qwen3-max-preview | 0.77 s Best: 0.55Worst: 1.56 | 44.58t/s | 5 |
| 19 | integrate.api.nvidia.comintegrate.api.nvidia.com | openai/gpt-oss-120b | 0.80 s Best: 0.52Worst: 1.30 | 251.35t/s | 5 |
| 20 | 1 123.54.215.139:8008123.54.215.139:8008 | qwen3-32b | 0.81 s Best: 0.29Worst: 2.11 | 22.93t/s | 5 |
| 21 | ETOS APIapi.ericterminal.com | moonshotai/kimi-k2-instruct-0905 | 0.83 s Best: 0.72Worst: 1.06 | 148.30t/s | 5 |
| 22 | 简 简小智API中转站newapi.jianxiaozhi.chat:56897 | deepseek-v3.2-exp | 0.85 s Best: 0.64Worst: 1.05 | 27.14t/s | 5 |
| 23 | 简小智API中转站newapi.jianxiaozhi.chat:56897 | deepseek-v3.2-exp | 0.85 s Best: 0.64Worst: 1.05 | 27.14t/s | 5 |
| 24 | DashScopedashscope.aliyuncs.com | qwen3-235b-a22b-instruct-2507 | 0.85 s Best: 0.55Worst: 1.46 | 52.41t/s | 10 |
| 25 | DashScopedashscope.aliyuncs.com | deepseek-v3 | 0.89 s Best: 0.69Worst: 1.21 | 35.73t/s | 5 |
| 26 | 8 82.157.254.69:300082.157.254.69:3000 | Qwen/Qwen3-32B | 0.91 s Best: 0.61Worst: 1.86 | 1214.07t/s | 5 |
| 27 | DashScopedashscope.aliyuncs.com | qwen-plus-2025-12-01 | 0.92 s Best: 0.63Worst: 1.23 | 52.41t/s | 5 |
| 28 | SiliconFlowapi.siliconflow.cn | Qwen/Qwen3-8B | 0.96 s Best: 0.65Worst: 1.73 | 20.90t/s | 5 |
| 29 | ChatGTPwww.chatgtp.cn | gemini-2.0-flash | 1.02 s Best: 0.81Worst: 1.60 | 181.28t/s | 5 |
| 30 | 简小智API中转站newapi.jianxiaozhi.chat:56897 | deepseek-v3-1-terminus | 1.03 s Best: 0.76Worst: 1.46 | 126.69t/s | 5 |
| 31 | 简 简小智API中转站newapi.jianxiaozhi.chat:56897 | deepseek-v3-1-terminus | 1.03 s Best: 0.76Worst: 1.46 | 126.69t/s | 5 |
| 32 | ChatGTPwww.chatgtp.cn | gpt-4.1-nano-2025-04-14 | 1.03 s Best: 0.71Worst: 1.32 | 592.40t/s | 10 |
| 33 | N New APIapi.seosycy.com | deepseek-v3-1-terminus | 1.05 s Best: 0.76Worst: 1.40 | 81.58t/s | 5 |
| 34 | A AI Toolsplatform.aitools.cfd | qwen/qwen2.5-7b | 1.13 s Best: 0.63Worst: 2.54 | 96.11t/s | 25 |
| 35 | a arkark.cn-beijing.volces.com | deepseek-v3-2-251201 | 1.24 s Best: 0.59Worst: 1.82 | 30.46t/s | 5 |
| 36 | DashScopedashscope.aliyuncs.com | deepseek-v3.2 | 1.25 s Best: 0.45Worst: 15.32 | 23.49t/s | 30 |
| 37 | A AI Toolsplatform.aitools.cfd | zhipu/glm-4-9b | 1.40 s Best: 0.84Worst: 2.13 | 62.28t/s | 5 |
| 38 | a ai-hub.square-llm.comai-hub.square-llm.com | google/gemini-3-flash-preview | 1.45 s Best: 0.93Worst: 1.83 | 150.13t/s | 5 |
| 39 | 全球AIglobalai.vip | gpt-oss-120b | 1.50 s Best: 1.27Worst: 1.74 | 284.17t/s | 5 |
| 40 | 8 82.157.254.69:300082.157.254.69:3000 | deepseek-r1 | 1.50 s Best: 1.17Worst: 2.07 | 56.00t/s | 5 |
| 41 | DashScopedashscope.aliyuncs.com | deepseek-v3.2-exp | 1.51 s Best: 0.55Worst: 4.58 | 29.09t/s | 5 |
| 42 | 黑与白公益站ai.hybgzs.com | gemini-2.5-flash-lite | 1.55 s Best: 0.85Worst: 2.40 | 186.31t/s | 5 |
| 43 | GPTAPI.USwww.gptapi.us | deepseek-chat | 1.56 s Best: 0.39Worst: 4.50 | 83.34t/s | 5 |
| 44 | A AI Toolsplatform.aitools.cfd | zhipu/glm-4-flash | 1.57 s Best: 1.02Worst: 2.05 | 33.00t/s | 5 |
| 45 | a api.xiaomimimo.comapi.xiaomimimo.com | mimo-v2-flash | 1.58 s Best: 0.72Worst: 4.06 | 102.45t/s | 5 |
| 46 | ChatGTPwww.chatgtp.cn | deepseek-v3-1-terminus | 1.65 s Best: 0.47Worst: 7.42 | 60.84t/s | 15 |
| 47 | A AI Toolsplatform.aitools.cfd | zhipu/glm-4-flash | 1.70 s Best: 0.45Worst: 20.71 | 32.23t/s | 450 |
| 48 | o ocool APIone.ocoolai.com | claude-sonnet-4-5-20250929 | 1.77 s Best: 1.29Worst: 2.32 | 45.03t/s | 5 |
| 49 | 算 算了么 APIapi.suanli.cn | zhipu/glm-4-flash | 1.78 s Best: 0.87Worst: 4.67 | 35.00t/s | 10 |
| 50 | N New APIapi.seosycy.com | gemini-2.0-flash | 1.79 s Best: 1.07Worst: 2.54 | 181.83t/s | 5 |