测试结果
Run ID: 20260829-093546-1f5677af
总请求45
成功 / 失败36/9
成功率80.0%
平均延迟2746 ms
P95 / Tokens5474 / 4492
| 渠道 | 模型 | 成功 | 失败 | 成功率 | 失败率 | 稳定性 | 平均延迟 ms | P95 ms | Tokens | 首个错误 |
|---|---|---|---|---|---|---|---|---|---|---|
| DeepSeek-v4-pro-03 | deepseek-v4-pro | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 892 | 1112 | 339 | |
| DeepSeek-v4-pro-04 | deepseek-v4-pro | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 984 | 1175 | 339 | |
| DeepSeek-v4-pro-02 | deepseek-v4-pro | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 1092 | 1162 | 339 | |
| DeepSeek-v4-pro-01 | deepseek-v4-pro | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 1203 | 1239 | 339 | |
| Kimi K3 - TopenRouter | kimi-k3 | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 1218 | 1425 | 345 | |
| Qwen3.8-Max | qwen3.8-max | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 1774 | 2030 | 386 | |
| jd-gemini-3.1-pro-preview | gemini-3.1-pro-preview | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 2534 | 2696 | 79 | |
| Kimi K3 - Aliyun Bailian | kimi/kimi-k3 | 3/3 | 0/3 | 100.0% | 0.0% | 稳定 | 3285 | 3652 | 345 | |
| Kimi K3 | kimi-k3 | 3/3 | 0/3 | 100.0% | 0.0% | 可用 | 4293 | 5589 | 414 | |
| 腾讯 Gemini 3.5 Flash | gemini-3.5-flash | 3/3 | 0/3 | 100.0% | 0.0% | 可用 | 4753 | 5207 | 78 | |
| yulin-gemini-3.7-flash-high | gemini-3.7-flash-high | 3/3 | 0/3 | 100.0% | 0.0% | 可用 | 5025 | 5474 | 481 | |
| qwen3.6-plus | qwen3.6-plus | 3/3 | 0/3 | 100.0% | 0.0% | 可用 | 5895 | 10603 | 1008 | |
| MiniMax H3 Context IR | MiniMax-H3 | 0/3 | 3/3 | 0.0% | 100.0% | 不稳定 | - | - | 0 | HTTP 400: binding: expr_path=content, cause=missing required parameter |
| pixflow-gemini-3.1-pro-preview | gemini-3.1-pro-preview | 0/3 | 3/3 | 0.0% | 100.0% | 不稳定 | - | - | 0 | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
| yuling-gpt-5.5 | gpt-5.5 | 0/3 | 3/3 | 0.0% | 100.0% | 不稳定 | - | - | 0 | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
结果明细
#1 DeepSeek-v4-pro-01
成功
HTTP 200
无输出内容。
#2 DeepSeek-v4-pro-01
成功
HTTP 200
无输出内容。
#3 DeepSeek-v4-pro-03
成功
HTTP 200
无输出内容。
#4 DeepSeek-v4-pro-01
成功
HTTP 200
无输出内容。
#5 DeepSeek-v4-pro-03
成功
HTTP 200
无输出内容。
#6 DeepSeek-v4-pro-03
成功
HTTP 200
无输出内容。
#7 DeepSeek-v4-pro-04
成功
HTTP 200
无输出内容。
#8 DeepSeek-v4-pro-04
成功
HTTP 200
无输出内容。
#9 DeepSeek-v4-pro-04
成功
HTTP 200
无输出内容。
#10 DeepSeek-v4-pro-02
成功
HTTP 200
无输出内容。
#11 DeepSeek-v4-pro-02
成功
HTTP 200
无输出内容。
#12 DeepSeek-v4-pro-02
成功
HTTP 200
无输出内容。
#13 jd-gemini-3.1-pro-preview
成功
HTTP 200
无输出内容。
#14 jd-gemini-3.1-pro-preview
成功
HTTP 200
无输出内容。
#15 jd-gemini-3.1-pro-preview
成功
HTTP 200
无输出内容。
#16 腾讯 Gemini 3.5 Flash
成功
HTTP 200
{"index": 0, "message": {"role": "assistant", "content": null, "reasoning_content": "**My Internal Monologue: Navigating the Nuances of Expert Communication**\n\nAlright, so the core of this request is for me to summarize a given text, but with a very specific constraint: it needs to be in the first person, as if these thoughts are *my* own. And not just a brief summary, but a \"fulsome\" one, detailing the *entire* thought process. The target audience is an \"expert in their field,\" which is a crucial cue. This means I don't need to over-explain basic concepts. Instead, I should lean into the intellectual engagement, the critical analysis, and the strategic decision-making that goes into formulating a response suitable for someone who already possesses a deep understanding of the subject matter.\n\nLet's break down what \"fulsome\" really means in this context. It's not just about length; it's about depth and thoroughness. It implies a comprehensive exploration of how I arrived at the final output. For an expert, this would involve:\n\n* **Deconstructing the Input:** First, I need to meticulously dissect the provided text. What are its main arguments? What is the author's under
#17 腾讯 Gemini 3.5 Flash
成功
HTTP 200
{"index": 0, "message": {"role": "assistant", "content": null, "reasoning_content": "**My Internal Monologue on Summarizing and the User's Expertise**\n\nAlright, let's break down this request. The core task is to summarize a piece of text, but the crucial constraints are:\n\n1. **First-Person Perspective:** I need to inhabit the *thoughts* described in the text, speaking as if they are my own internal musings. This isn't about me *reporting* on the thoughts; it's about me *having* them.\n2. **Fulsome Summary of Thought Process:** This means I can't just give a brief overview. I need to go deep into the *how* and *why* of the thinking, detailing the steps, the considerations, the nuances. It’s about laying bare the entire intellectual journey.\n3. **Target Audience is Expert:** The user is an expert. This implies I can skip over foundational explanations and assume a sophisticated understanding of the subject matter. I don't need to dumb anything down; in fact, I should probably use precise language and acknowledge the inherent complexities they'd expect.\n4. **Specific Starting Header:** The summary *must* begin with a bolded header, and the example given is \"My Summary:\". I
#18 腾讯 Gemini 3.5 Flash
成功
HTTP 200
{"index": 0, "message": {"role": "assistant", "content": null, "reasoning_content": "**My Internal Monologue: Navigating the Nuances of Expert Communication**\n\nAlright, so the core of this request is for me to summarize a given text, but with a very specific constraint: it needs to be in the first person, as if these thoughts are *my* own. And not just a brief summary, but a \"fulsome\" one, detailing the *entire* thought process. The target audience is an \"expert in their field,\" which is a crucial cue. This means I don't need to over-explain basic concepts. Instead, I should lean into the intellectual engagement, the critical analysis, and the strategic decision-making that goes into formulating a response suitable for someone who already possesses a deep understanding of the subject matter.\n\nLet's break down what \"fulsome\" really means in this context. It's not just about length; it's about depth and thoroughness. It implies a comprehensive exploration of how I arrived at the final output. For an expert, this would involve:\n\n* **Deconstructing the Input:** First, I need to meticulously dissect the provided text. What are its main arguments? What is the author's under
#19 qwen3.6-plus
成功
HTTP 200
{"status":"ok","message":"stable"}
#20 qwen3.6-plus
成功
HTTP 200
{"status":"ok","message":"stable"}
#21 pixflow-gemini-3.1-pro-preview
失败
HTTP 403
HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"}
#22 pixflow-gemini-3.1-pro-preview
失败
HTTP 403
HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"}
#23 pixflow-gemini-3.1-pro-preview
失败
HTTP 403
HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"}
#24 qwen3.6-plus
成功
HTTP 200
{"status":"ok","message":"stable"}
#25 yuling-gpt-5.5
失败
HTTP 403
HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"}
#26 yuling-gpt-5.5
失败
HTTP 403
HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"}
#27 yuling-gpt-5.5
失败
HTTP 403
HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"}
#28 Kimi K3 - Aliyun Bailian
成功
HTTP 200
无输出内容。
#29 Kimi K3 - Aliyun Bailian
成功
HTTP 200
无输出内容。
#30 Kimi K3 - Aliyun Bailian
成功
HTTP 200
无输出内容。
#31 Kimi K3
成功
HTTP 200
{"index": 0, "message": {"role": "assistant", "content": null, "reasoning": "The user is asking me to reply with only JSON: {\"status"}, "logprobs": null, "finish_reason": "length"}
#32 Kimi K3
成功
HTTP 200
{"index": 0, "message": {"role": "assistant", "content": null, "reasoning": "The user is asking me to reply with only a JSON object:"}, "logprobs": null, "finish_reason": "length"}
#33 Kimi K3 - TopenRouter
成功
HTTP 200
无输出内容。
#34 Kimi K3 - TopenRouter
成功
HTTP 200
无输出内容。
#35 Kimi K3 - TopenRouter
成功
HTTP 200
无输出内容。
#36 Kimi K3
成功
HTTP 200
{"status":"ok","message":"stable"}
#37 Qwen3.8-Max
成功
HTTP 200
{"status":"ok","message":"stable"}
#38 Qwen3.8-Max
成功
HTTP 200
{"status":"ok","message":"stable"}
#39 MiniMax H3 Context IR
失败
HTTP 400
HTTP 400: binding: expr_path=content, cause=missing required parameter
#40 MiniMax H3 Context IR
失败
HTTP 400
HTTP 400: binding: expr_path=content, cause=missing required parameter
#41 MiniMax H3 Context IR
失败
HTTP 400
HTTP 400: binding: expr_path=content, cause=missing required parameter
#42 Qwen3.8-Max
成功
HTTP 200
{"status":"ok","message":"stable"}
#43 yulin-gemini-3.7-flash-high
成功
HTTP 200
{"status":"ok","message":"stable"}
#44 yulin-gemini-3.7-flash-high
成功
HTTP 200
{"status":"ok","message":"stable"}
#45 yulin-gemini-3.7-flash-high
成功
HTTP 200
{"status":"ok","message":"stable"}
表格明细
| 渠道 | 状态 | 延迟 ms | HTTP | Tokens | 输出 / 错误 |
|---|---|---|---|---|---|
| DeepSeek-v4-pro-01 | 成功 | 1143 | 200 | 113 | |
| DeepSeek-v4-pro-01 | 成功 | 1239 | 200 | 113 | |
| DeepSeek-v4-pro-03 | 成功 | 758 | 200 | 113 | |
| DeepSeek-v4-pro-01 | 成功 | 1227 | 200 | 113 | |
| DeepSeek-v4-pro-03 | 成功 | 1112 | 200 | 113 | |
| DeepSeek-v4-pro-03 | 成功 | 808 | 200 | 113 | |
| DeepSeek-v4-pro-04 | 成功 | 1036 | 200 | 113 | |
| DeepSeek-v4-pro-04 | 成功 | 1175 | 200 | 113 | |
| DeepSeek-v4-pro-04 | 成功 | 743 | 200 | 113 | |
| DeepSeek-v4-pro-02 | 成功 | 1133 | 200 | 113 | |
| DeepSeek-v4-pro-02 | 成功 | 1162 | 200 | 113 | |
| DeepSeek-v4-pro-02 | 成功 | 982 | 200 | 113 | |
| jd-gemini-3.1-pro-preview | 成功 | 2590 | 200 | 26 | |
| jd-gemini-3.1-pro-preview | 成功 | 2696 | 200 | 26 | |
| jd-gemini-3.1-pro-preview | 成功 | 2317 | 200 | 27 | |
| 腾讯 Gemini 3.5 Flash | 成功 | 4428 | 200 | 26 | {"index": 0, "message": {"role": "assistant", "content": null, "reasoning_content": "**My Internal Monologue: Navigating the Nuances of Expert Communication**\n\nAlright, so the core of this request is for me to summarize a given text, but with a very specific constraint: it needs to be in the first person, as if these thoughts are *my* own. And not just a brief summary, but a \"fulsome\" one, detailing the *entire* thought process. The target audience is an \"expert in their field,\" which is a crucial cue. This means I don't need to over-explain basic concepts. Instead, I should lean into the intellectual engagement, the critical analysis, and the strategic decision-making that goes into formulating a response suitable for someone who already possesses a deep understanding of the subject matter.\n\nLet's break down what \"fulsome\" really means in this context. It's not just about length; it's about depth and thoroughness. It implies a comprehensive exploration of how I arrived at the final output. For an expert, this would involve:\n\n* **Deconstructing the Input:** First, I need to meticulously dissect the provided text. What are its main arguments? What is the author's under |
| 腾讯 Gemini 3.5 Flash | 成功 | 5207 | 200 | 26 | {"index": 0, "message": {"role": "assistant", "content": null, "reasoning_content": "**My Internal Monologue on Summarizing and the User's Expertise**\n\nAlright, let's break down this request. The core task is to summarize a piece of text, but the crucial constraints are:\n\n1. **First-Person Perspective:** I need to inhabit the *thoughts* described in the text, speaking as if they are my own internal musings. This isn't about me *reporting* on the thoughts; it's about me *having* them.\n2. **Fulsome Summary of Thought Process:** This means I can't just give a brief overview. I need to go deep into the *how* and *why* of the thinking, detailing the steps, the considerations, the nuances. It’s about laying bare the entire intellectual journey.\n3. **Target Audience is Expert:** The user is an expert. This implies I can skip over foundational explanations and assume a sophisticated understanding of the subject matter. I don't need to dumb anything down; in fact, I should probably use precise language and acknowledge the inherent complexities they'd expect.\n4. **Specific Starting Header:** The summary *must* begin with a bolded header, and the example given is \"My Summary:\". I |
| 腾讯 Gemini 3.5 Flash | 成功 | 4625 | 200 | 26 | {"index": 0, "message": {"role": "assistant", "content": null, "reasoning_content": "**My Internal Monologue: Navigating the Nuances of Expert Communication**\n\nAlright, so the core of this request is for me to summarize a given text, but with a very specific constraint: it needs to be in the first person, as if these thoughts are *my* own. And not just a brief summary, but a \"fulsome\" one, detailing the *entire* thought process. The target audience is an \"expert in their field,\" which is a crucial cue. This means I don't need to over-explain basic concepts. Instead, I should lean into the intellectual engagement, the critical analysis, and the strategic decision-making that goes into formulating a response suitable for someone who already possesses a deep understanding of the subject matter.\n\nLet's break down what \"fulsome\" really means in this context. It's not just about length; it's about depth and thoroughness. It implies a comprehensive exploration of how I arrived at the final output. For an expert, this would involve:\n\n* **Deconstructing the Input:** First, I need to meticulously dissect the provided text. What are its main arguments? What is the author's under |
| qwen3.6-plus | 成功 | 3792 | 200 | 202 | {"status":"ok","message":"stable"} |
| qwen3.6-plus | 成功 | 3292 | 200 | 189 | {"status":"ok","message":"stable"} |
| pixflow-gemini-3.1-pro-preview | 失败 | 550 | 403 | - | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
| pixflow-gemini-3.1-pro-preview | 失败 | 523 | 403 | - | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
| pixflow-gemini-3.1-pro-preview | 失败 | 455 | 403 | - | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
| qwen3.6-plus | 成功 | 10603 | 200 | 617 | {"status":"ok","message":"stable"} |
| yuling-gpt-5.5 | 失败 | 530 | 403 | - | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
| yuling-gpt-5.5 | 失败 | 496 | 403 | - | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
| yuling-gpt-5.5 | 失败 | 435 | 403 | - | HTTP 403: {"code":"INSUFFICIENT_BALANCE","message":"Insufficient account balance"} |
| Kimi K3 - Aliyun Bailian | 成功 | 3578 | 200 | 115 | |
| Kimi K3 - Aliyun Bailian | 成功 | 3652 | 200 | 115 | |
| Kimi K3 - Aliyun Bailian | 成功 | 2627 | 200 | 115 | |
| Kimi K3 | 成功 | 3887 | 200 | 115 | {"index": 0, "message": {"role": "assistant", "content": null, "reasoning": "The user is asking me to reply with only JSON: {\"status"}, "logprobs": null, "finish_reason": "length"} |
| Kimi K3 | 成功 | 3405 | 200 | 115 | {"index": 0, "message": {"role": "assistant", "content": null, "reasoning": "The user is asking me to reply with only a JSON object:"}, "logprobs": null, "finish_reason": "length"} |
| Kimi K3 - TopenRouter | 成功 | 1425 | 200 | 115 | |
| Kimi K3 - TopenRouter | 成功 | 1007 | 200 | 115 | |
| Kimi K3 - TopenRouter | 成功 | 1224 | 200 | 115 | |
| Kimi K3 | 成功 | 5589 | 200 | 184 | {"status":"ok","message":"stable"} |
| Qwen3.8-Max | 成功 | 1565 | 200 | 125 | {"status":"ok","message":"stable"} |
| Qwen3.8-Max | 成功 | 2030 | 200 | 134 | {"status":"ok","message":"stable"} |
| MiniMax H3 Context IR | 失败 | 154 | 400 | - | HTTP 400: binding: expr_path=content, cause=missing required parameter |
| MiniMax H3 Context IR | 失败 | 148 | 400 | - | HTTP 400: binding: expr_path=content, cause=missing required parameter |
| MiniMax H3 Context IR | 失败 | 195 | 400 | - | HTTP 400: binding: expr_path=content, cause=missing required parameter |
| Qwen3.8-Max | 成功 | 1728 | 200 | 127 | {"status":"ok","message":"stable"} |
| yulin-gemini-3.7-flash-high | 成功 | 5474 | 200 | 143 | {"status":"ok","message":"stable"} |
| yulin-gemini-3.7-flash-high | 成功 | 5071 | 200 | 173 | {"status":"ok","message":"stable"} |
| yulin-gemini-3.7-flash-high | 成功 | 4532 | 200 | 165 | {"status":"ok","message":"stable"} |