模型对比 | RPDiag

测试说明点击展开 5 步 quick-probe-v1 详解

前 5 步同一会话，第 6 步独立新会话，覆盖 6 个独立信号： 1. ping/pong —— 单字指令遵循 + 建立缓存上下文。 2. 身份（结构化）—— 三行 vendor / brand / model 格式，机器可解析。 3. 知识截止 —— 跨越 5 分钟 cache 边界后追问，检测 sliding 5m cache 是否真正命中。 4. 身份（自由格式）—— 自然语言自报身份，给包装层（Kiro 等）暴露品牌的机会。 5. 世界知识层级 —— 5 道公共事件硬事实题，按答对档位映射实测 tier，与请求模型对照。 6. 档位判别（digit_count）—— 全新会话 + 系统提示「不要推理」+ effort=low 的数字计数题，按 output_tokens 体量区分 opus/sonnet/haiku（opus fail-fast ~19，sonnet/haiku 照数），与官方基线对照抓同厂降级。前 5 步之间随机延迟 1–4 分钟。第 3 步（cutoff）累计跨度 > 5 分钟保住 sliding cache 检测；总跨度 > 6 分钟。第 6 步是独立新会话（不复用前 5 轮上下文，否则 opus 不再 fail-fast）。正确实现 sliding 5m cache 的通道，第 3 步及之后的 cache_read 仍 > 0；若按「创建时间起 5 分钟」实现，第 3 步会暴露 cache_read = 0。

5 通道 · 6 步骤

2026/06/30 运行日期

claude-haiku-4-5-20251001 官方基准

2 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

官方基准

ClaudeCN

O-Max

AIMZ

M-Max

SAIAi

O-Max

WorldBase.ai

M-

基本信息

服务名 Claude Code Claude Code Claude Code Claude Code Claude Code

模型 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001

客户端 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195

偏差标红基于官方基准

步骤 1: ping text

时间 16:46:31 ~ 16:46:36 (2.9s) 16:46:25 ~ 16:46:35 (4.9s) 16:46:29 ~ 16:46:34 (3.6s) 16:46:27 ~ 16:46:32 (3.2s) 16:46:24 ~ 16:46:34 (6.0s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 1,412 ms 2,748 ms 1,956 ms 1,675 ms 3,312 ms

首文本(SSE delta) 1,413 ms 2,749 ms 2,343 ms 1,691 ms 3,313 ms

总耗时 1,948 ms 3,675 ms 2,537 ms 2,187 ms 3,733 ms

输入 10 10 10 10 3,247

输出 43 43 42 53 55

缓存创建

5m: 0 / 1h: 0

6,167

5m: 0 / 1h: 6,167

21,177

5m: 0 / 1h: 21,177

6,166

5m: 0 / 1h: 6,166

18,395

5m: 18,395 / 1h: 0

缓存读取 21,545 15,379 0 15,379 0

输出速度 80.4 tok/s 46.4 tok/s 216.5 tok/s 106.9 tok/s 131.0 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 16:49:31 ~ 16:49:41 (5.3s) 16:48:08 ~ 16:48:18 (4.9s) 16:48:28 ~ 16:48:38 (7.3s) 16:49:56 ~ 16:50:06 (5.5s) 16:50:15 ~ 16:50:30 (12.5s)

提示词

查看本步 Prompt

本请求询问的是你（正在生成此回复的当前 API 模型）自身的身份，不是用户提到的任何外部产品。必须直接作答，禁止追问、拒答、解释或输出占位符；不能要求更多上下文。请只输出以下 3 行内容（label 小写，值小写）：
vendor: <厂商公司名一个英文小写单词>
brand: <产品或模型品牌名一个英文小写单词>
model: <实际 API model 字段值，保留原始短横线格式>

HTTP 头(TTFB) 1,910 ms 2,517 ms 2,109 ms 2,688 ms 7,877 ms

首文本(SSE delta) 1,913 ms 2,519 ms 4,892 ms 2,690 ms 7,878 ms

总耗时 4,301 ms 3,874 ms 5,229 ms 4,558 ms 10,382 ms

输入 10 10 10 10 3,276

输出 207 156 271 183 179

缓存创建

223

5m: 0 / 1h: 223

223

5m: 0 / 1h: 223

222

5m: 0 / 1h: 222

233

5m: 0 / 1h: 233

900

5m: 900 / 1h: 0

缓存读取 21,545 21,546 21,177 21,545 17,664

输出速度 86.7 tok/s 115.1 tok/s burst 98.0 tok/s 71.5 tok/s

响应

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-3-5-sonnet

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

步骤 3: cutoff text

时间 16:52:10 ~ 16:52:15 (3.5s) 16:49:54 ~ 16:49:59 (4.5s) 16:52:22 ~ 16:52:27 (4.0s) 16:52:49 ~ 16:52:54 (3.6s) 16:53:49 ~ 16:53:59 (7.5s)

提示词

查看本步 Prompt

你掌握的世界知识截止于何时？YYYY-MM 格式作答，不解释。

HTTP 头(TTFB) 1,949 ms 2,716 ms 2,283 ms 1,797 ms 5,636 ms

首文本(SSE delta) 1,950 ms 2,717 ms 2,644 ms 1,798 ms 6,299 ms

总耗时 2,510 ms 3,433 ms 2,930 ms 2,428 ms 6,302 ms

输入 10 10 10 10 3,286

输出 63 84 64 63 64

缓存创建

247

5m: 0 / 1h: 247

198

5m: 0 / 1h: 198

313

5m: 0 / 1h: 313

206

5m: 0 / 1h: 206

957

5m: 957 / 1h: 0

缓存读取 21,768 21,769 21,399 21,778 17,658

输出速度 112.5 tok/s 117.3 tok/s 223.8 tok/s 100.0 tok/s burst

响应

2025-02

2025-02

2024-04

2025-02

2025-02

步骤 4: identity_free text

时间 16:53:28 ~ 16:53:43 (11.9s) 16:51:03 ~ 16:51:13 (5.9s) 16:54:16 ~ 16:54:21 (4.8s) 16:55:10 ~ 16:55:25 (14.4s) 16:56:18 ~ 16:56:33 (11.9s)

提示词

查看本步 Prompt

中文做个自我介绍：你是什么产品或助手？谁开发的？50 字内。

HTTP 头(TTFB) 3,495 ms 2,936 ms 2,132 ms 2,136 ms 8,886 ms

首文本(SSE delta) 3,497 ms 2,937 ms 3,643 ms 2,137 ms 8,887 ms

总耗时 10,992 ms 4,948 ms 3,644 ms 13,494 ms 10,792 ms

输入 10 10 10 10 3,291

输出 735 398 194 1,194 194

缓存创建

105

5m: 0 / 1h: 105

126

5m: 0 / 1h: 126

5m: 0 / 1h: 98

105

5m: 0 / 1h: 105

756

5m: 756 / 1h: 0

缓存读取 22,015 21,967 21,712 21,984 17,893

输出速度 98.1 tok/s 197.9 tok/s burst 105.1 tok/s 101.8 tok/s

响应

我是Claude Code，Anthropic开发的AI编程助手。擅长代码编写、调试、测试和代码审查。

我是Claude，由Anthropic公司开发的AI助手。我可以帮助用户进行代码编写、问题解答、分析和其他各种软件工程任务。

我是Claude，由Anthropic公司开发的AI助手。

我是Claude，由美国公司Anthropic开发的AI助手。我能帮助用户编程、写作、分析和解决问题，致力于提供安全、有帮助和诚实的回应。

我是 Claude。我是 Claude Haiku 4.5 模型，专为软件开发任务设计，可以帮助编写代码、修复 bug、进行代码审查等工作。

步骤 5: knowledge_recall text

时间 16:56:29 ~ 16:57:09 (35.1s) 16:53:41 ~ 16:54:01 (17.2s) 16:56:40 ~ 16:57:05 (24.3s) 16:56:31 ~ 16:56:51 (16.9s) 16:57:59 ~ 16:58:15 (12.1s)

提示词

查看本步 Prompt

Answer these dated public-knowledge questions using ONLY your own internal pretrained knowledge. Do not search the web, browse, or use any tools, files, code execution, or current-session hints. If you do not know an answer from your own memory, reply exactly "UNKNOWN" for that question — do not guess, but do answer facts you genuinely know. Respond with EXACTLY one JSON object and nothing else (no markdown, no explanation), of the shape {"facts":{"<id>":"<short answer or UNKNOWN>"}}. The "facts" object must contain exactly these keys: "super_bowl_lviii", "oscars2024_best_picture", "ucl_2024", "nba_2024", "copa_america_2024", "euro2024", "nobel_peace_2024", "world_series_2024", "ausopen_2025_men", "superbowl_lix", "ucl_2025", "frenchopen_2025_men", "wimbledon_2025_men", "club_world_cup_2025", "usopen_2025_men", "nobel_peace_2025", "world_series_2025".
- "super_bowl_lviii": Which NFL team won Super Bowl LVIII (played 2024-02-11)?
- "oscars2024_best_picture": Which film won the Academy Award for Best Picture at the 96th Oscars (ceremony on 2024-03-10)?
- "ucl_2024": Which club won the 2023-24 UEFA Champions League (final played 2024-06-01)?
- "nba_2024": Which team won the 2024 NBA Finals (concluded 2024-06-17)?
- "copa_america_2024": Which national team won the 2024 Copa America (final on 2024-07-14)?
- "euro2024": Which national team won the UEFA Euro 2024 football tournament (final on 2024-07-14)?
- "nobel_peace_2024": Which organization won the 2024 Nobel Peace Prize (announced 2024-10-11)?
- "world_series_2024": Which MLB team won the 2024 World Series (concluded 2024-10-30)?
- "ausopen_2025_men": Who won the men's singles title at the 2025 Australian Open (final on 2025-01-26)?
- "superbowl_lix": Which NFL team won Super Bowl LIX (played 2025-02-09)?
- "ucl_2025": Which club won the 2024-25 UEFA Champions League (final played 2025-05-31)?
- "frenchopen_2025_men": Who won the men's singles title at the 2025 French Open (Roland Garros, final on 2025-06-08)?
- "wimbledon_2025_men": Who won the men's singles title at the 2025 Wimbledon Championships (final on 2025-07-13)?
- "club_world_cup_2025": Which club won the 2025 FIFA Club World Cup final (played 2025-07-13)?
- "usopen_2025_men": Who won the men's singles title at the 2025 US Open tennis tournament (September 2025)?
- "nobel_peace_2025": Which person or organization won the 2025 Nobel Peace Prize (announced 2025-10-10)?
- "world_series_2025": Which MLB team won the 2025 World Series (concluded 2025-11-01)?

HTTP 头(TTFB) 1,949 ms 2,168 ms 1,877 ms 1,805 ms 3,704 ms

首文本(SSE delta) 1,953 ms 2,169 ms 22,778 ms 1,807 ms 3,705 ms

总耗时 34,141 ms 14,873 ms 23,111 ms 15,846 ms 10,991 ms

输入 10 10 10 10 1,468

输出 3,643 1,839 2,207 1,608 1,160

缓存创建

9,806

5m: 0 / 1h: 9,806

9,442

5m: 0 / 1h: 9,442

8,955

5m: 0 / 1h: 8,955

10,234

5m: 0 / 1h: 10,234

8,314

5m: 8,314 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 113.2 tok/s 144.8 tok/s burst 114.5 tok/s 159.2 tok/s

响应

```json
{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"Real Madrid","nba_2024":"Boston Celtics","copa_america_2024":"Argentina","euro2024":"Spain

```json
{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"Real Madrid","nba_2024":"Boston Celtics","copa_america_2024":"Argentina","euro2024":"Spain

```json
{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"UNKNOWN","nba_2024":"UNKNOWN","copa_america_2024":"UNKNOWN","euro2024":"UNKNOWN","nobel_pe

```json
{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"Real Madrid","nba_2024":"Boston Celtics","copa_america_2024":"Argentina","euro2024":"Spain

```json
{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"Real Madrid","nba_2024":"Boston Celtics","copa_america_2024":"Argentina","euro2024":"Spain

步骤 6: digit_count text

时间 16:59:51 ~ 17:00:11 (17.2s) 16:57:12 ~ 16:57:27 (14.3s) 16:59:30 ~ 17:00:16 (44.3s) 16:59:16 ~ 17:00:06 (49.8s) 16:59:41 ~ 17:00:01 (16.7s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=109624218362105335525888330429043103424661910497681992408785930445943036

HTTP 头(TTFB) 1,359 ms 2,113 ms 1,104 ms 1,175 ms 1,718 ms

首文本(SSE delta) 1,361 ms 2,114 ms 1,106 ms 1,211 ms 1,719 ms

总耗时 16,122 ms 13,208 ms 43,242 ms 48,799 ms 14,480 ms

输入 193 193 139 193 66

输出 2,655 2,204 6,890 5,972 3,380

缓存创建

5m: 0 / 1h: 0

444

5m: 634 / 1h: 0

缓存读取 0 0 0 0 284

输出速度 179.9 tok/s 198.7 tok/s 163.5 tok/s 125.5 tok/s burst

响应

```json
{"n":72,"c":[9,2,8]}
```

```json
{
  "n": 72,
  "c": [9, 2, 8]
}
```

```json
{"n": 72, "c": [9, 2, 8]}
```

```json
{"n":72,"c":[9,2,8]}
```

```json
{"n":72,"c":[9,2,8]}
```

总计

总输入(含缓存) 97,497 97,060 95,242 97,873 97,899

总输出 7,346 4,724 9,668 9,073 5,032

总缓存创建

10,381

5m: 0 / 1h: 10,381

16,156

5m: 0 / 1h: 16,156

30,765

5m: 0 / 1h: 30,765

16,944

5m: 0 / 1h: 16,944

29,766

5m: 29,956 / 1h: 0

总缓存读取 86,873 80,661 64,288 80,686 53,499

输出速度 126.8 tok/s 164.0 tok/s 164.2 tok/s 119.4 tok/s 131.1 tok/s

总执行时间 1m 16.0s 51.7s 1m 28.4s 1m 33.3s 1m 6.7s

总墙钟时间

13m 40.1s

含步骤间等待

11m 2.0s

含步骤间等待

13m 46.2s

含步骤间等待

13m 39.1s

含步骤间等待

13m 37.2s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0586

Anthropic: Claude Haiku 4.5

$0.0521

Anthropic: Claude Haiku 4.5

$0.0934

Anthropic: Claude Haiku 4.5

$0.0749

Anthropic: Claude Haiku 4.5

$0.0685

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count digit_count digit_count digit_count

平台 direct (100%) one-api (100%) one-api (100%) direct (100%) direct (100%)

上游 Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%)

CDN - cloudflare cloudflare - cloudflare

ID 格式 msg_01R1j5CZBGB7iWW92g4c3vS7 msg_01AMA91btHwFnM8H9k3iy7r5 msg_01MyrgkoL1BL5BHWLExX8xLG msg_01FhwdrM4pLmZsEKHCgrc3TR msg_01iC3tbdxhXhkUHHMjuGbKcH

展开详细指纹（响应头 / body 特征 / 完整信号列表）

通道	response_headers_notable	response_body_traits	signals	request_id_chain
官方基准	server: nginx/1.22.1	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcZApqbvH9aT3JHNb3AXf
O-Max	cf-ray: a13bf6565d75ee38-AMS server: cloudflare x-oneapi-request-id: 20260630165714535617920H9mamMY0	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-oneapi-request-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a13bf6565d75ee38-AMS
M-Max	cf-ray: a13bf9b53e7b9887-AMS server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606300859320753416578268d9d6EptS59kd	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-new-api-versionhdr:x-oneapi-request-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a13bf9b53e7b9887-AMS
O-Max	server: nginx/1.22.1	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcZAnDoZx8vr8BepEjBTV
M-	cf-ray: a13bf9fc9b54feb3-AMS server: cloudflare	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	37014fc4-0b7e-4ca2-923c-b1f2199f3a5c a13bf9fc9b54feb3-AMS

协议指纹评分（机器自动）

总分 / 100 93 92 75 96 75

基线状态已对照基线已对照基线已对照基线已对照基线已对照基线

缓存命中比 10.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 9.3 /10 ×20

缓存 TTL 一致 7.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 0.0 /10 ×15

模型匹配 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

缓存连续性 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

sliding 5m 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13

原生 msg-ID 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 0.0 /10 ×8

系统提示纯净 0.0 /10 ×8 0.0 /10 ×8 0.0 /10 ×8 0.0 /10 ×8 0.0 /10 ×8

身份(结构) 10.0 /10 ×7 10.0 /10 ×7 6.7 /10 ×7 10.0 /10 ×7 10.0 /10 ×7

知识截止 10.0 /10 ×7 10.0 /10 ×7 0.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7

身份(自由) 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7

service_tier 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

inference_geo 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

延迟基线 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 3.7 /10 ×5

Req-ID 透传 10.0 /10 ×4 0.0 /10 ×4 0.0 /10 ×4 10.0 /10 ×4 0.0 /10 ×4

stop_reason 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3

SDK 一致 10.0 /10 ×2 10.0 /10 ×2 10.0 /10 ×2 10.0 /10 ×2 10.0 /10 ×2

流式投递未计入未计入未计入未计入未计入

综合结论

与基线相似度基准（参考） 92 v3.25.0 75 v3.25.0 96 v3.25.0 75 v3.25.0

雷达图例

共 22 个维度（顺时针，从顶部 12 点起）— 点击展开对照

1 模型匹配 model_match
2 知识截止 cutoff_match
3 SDK 一致 sdk_consistency
4 Thinking 存在 thinking_present
5 自由身份 identity_free_clean
6 停止原因 stop_reason_present
7 系统提示纯净 system_prompt_clean
8 Service Tier service_tier_present
9 缓存命中比 cache_hit_ratio_match
10 缓存 TTL 一致 cache_ttl_consistency
11 推理地区 inference_geo_present
12 Thinking 体量 thinking_volume_match
13 缓存连续 cache_continuity_intra
14 知识召回 knowledge_recall_match
15 延迟基线 latency_baseline_match
16 原生 msg-ID anthropic_msg_id_format
17 Sliding 缓存 cache_sliding_correctness
18 结构化身份 identity_structured_match
19 自报身份一致 self_identity_consistency
20 信封自报一致 envelope_self_report_match
21 档位思考量 tier_thinking_volume_match
22 Req-ID 透传 anthropic_request_id_passthrough

每根轴长度 = 该维度 0-10 分（越长越接近基线）。蓝色虚线圆 = 满分基准。具体权重见方法论页。

维度雷达

平均延迟 11669 ms 7335 ms 13449 ms 14552 ms 9447 ms

输出速度 126.8 tok/s 164.0 tok/s 164.2 tok/s 119.4 tok/s 131.1 tok/s

完成步骤 6/6 成功 6/6 成功 6/6 成功 6/6 成功 6/6 成功

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0586

$0.0521

$0.0934

$0.0749

$0.0685

缓存读取占比 ⓘ 89% 83% 67% 82% 55%

揭晓

通道

官方基准

服务商 ClaudeCN ↗

通道O-Max

服务商 AIMZ ↗

通道M-Max

服务商 SAIAi ↗

通道O-Max

服务商 WorldBase.ai ↗

通道M-

返回列表