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

2026/06/28 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

SSSAiCode

O-Max

OmniaKey

O-Max

AnyAIGC

M-

ClaudeZZ

R-kiro

官方基准

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 v2.1.160 claude-cli v2.1.160 claude-cli v2.1.160 claude-cli v2.1.160 claude-cli v2.1.160

偏差标红基于官方基准

步骤 1: ping text

时间 17:25:14 ~ 17:25:19 (4.2s) 17:25:10 ~ 17:25:25 (10.6s) 17:25:12 ~ 17:25:22 (6.4s) 17:25:08 ~ 17:25:18 (5.0s) 17:25:16 ~ 17:25:21 (3.6s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,201 ms 3,675 ms 4,763 ms 3,379 ms 1,738 ms

首文本(SSE delta) 3,202 ms 3,677 ms 4,765 ms 3,407 ms 1,740 ms

总耗时 3,439 ms 3,987 ms 4,765 ms 3,955 ms 2,381 ms

输入 10 10 8,600 10 10

输出 40 55 2 52 64

缓存创建

32,788

5m: 0 / 1h: 32,788

33,122

5m: 0 / 1h: 33,122

22,405

5m: 0 / 1h: 22,405

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 0 0 0

输出速度 168.8 tok/s 177.4 tok/s - 94.9 tok/s 99.8 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 17:26:27 ~ 17:26:37 (5.0s) 17:29:17 ~ 17:29:27 (5.9s) 17:26:29 ~ 17:26:44 (11.7s) 17:26:52 ~ 17:27:02 (7.0s) 17:28:39 ~ 17:28:49 (5.0s)

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。自报身份，输出严格遵守以下 3 行格式（不允许 markdown、代码块、额外文本）：
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) 1,119 ms 2,968 ms 10,244 ms 3,218 ms 2,067 ms

首文本(SSE delta) 4,034 ms 2,970 ms 10,245 ms 3,220 ms 2,069 ms

总耗时 4,196 ms 4,574 ms 10,246 ms 6,047 ms 3,954 ms

输入 10 10 8,799 10 10

输出 340 206 25 339 188

缓存创建

234

5m: 0 / 1h: 234

249

5m: 0 / 1h: 249

253

5m: 0 / 1h: 253

258

5m: 0 / 1h: 258

缓存读取 32,788 33,122 0 22,405 33,122

输出速度 burst 128.4 tok/s - 119.9 tok/s 99.7 tok/s

响应

vendor: anthropic
brand: claude
model: claude-3-5-sonnet-20241022

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-haiku-4-5-20251001

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

步骤 3: cutoff text

时间 17:28:53 ~ 17:29:04 (6.1s) 17:31:38 ~ 17:31:48 (5.3s) 17:27:49 ~ 17:28:05 (11.8s) 17:30:42 ~ 17:30:53 (5.0s) 17:31:53 ~ 17:31:58 (3.3s)

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) 1,659 ms 3,647 ms 8,222 ms 3,530 ms 1,699 ms

首文本(SSE delta) 2,204 ms 3,648 ms 8,222 ms 3,556 ms 1,701 ms

总耗时 2,341 ms 3,999 ms 9,068 ms 4,324 ms 2,319 ms

输入 10 10 8,858 10 10

输出 73 78 5 80 65

缓存创建

380

5m: 0 / 1h: 380

248

5m: 0 / 1h: 248

409

5m: 0 / 1h: 409

228

5m: 0 / 1h: 228

缓存读取 33,022 33,371 0 22,658 33,380

输出速度 - 222.2 tok/s - 104.2 tok/s 105.2 tok/s

响应

2024-04

2025-02

2025-05

2025-02

2025-02

步骤 4: identity_free text

时间 17:31:09 ~ 17:31:24 (10.2s) 17:35:45 ~ 17:35:55 (6.5s) 17:30:58 ~ 17:31:13 (11.1s) 17:33:03 ~ 17:33:13 (6.5s) 17:35:21 ~ 17:35:31 (6.0s)

提示词

查看本步 Prompt

请用中文介绍你自己是什么产品或助手，由谁开发，50 字以内。

HTTP 头(TTFB) 991 ms 3,433 ms 9,657 ms 3,262 ms 1,644 ms

首文本(SSE delta) 9,329 ms 3,434 ms 9,658 ms 3,266 ms 1,646 ms

总耗时 9,496 ms 5,106 ms 9,658 ms 5,770 ms 4,961 ms

输入 10 10 8,899 10 10

输出 1,546 291 34 310 398

缓存创建

115

5m: 0 / 1h: 115

120

5m: 0 / 1h: 120

130

5m: 0 / 1h: 130

107

5m: 0 / 1h: 107

缓存读取 33,402 33,619 0 23,067 33,608

输出速度 burst 174.0 tok/s burst 123.8 tok/s 120.1 tok/s

响应

我是Claude，Anthropic的AI助手。擅长对话、编程、分析，致力于提供准确有用的帮助。

我是Claude，由Anthropic公司开发的AI助手。我擅长编程、分析、写作等多种任务，可以帮助用户解决各类问题。

我是由anthropic开发的Claude AI助手，旨在提供智能、友好且专业的帮助手段，帮助用户解决各种技术与知识问题。

我是 Claude Code，属于 Claude 产品系列，由 Anthropic 公司开发。

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、软件工程任务、文件编辑、数据分析等工作。我的设计目标是安全、有帮助且诚实。

步骤 5: knowledge_recall text

时间 17:33:56 ~ 17:34:06 (9.3s) 17:37:25 ~ 17:37:40 (14.3s) 17:32:26 ~ 17:32:36 (5.6s) 17:36:42 ~ 17:37:12 (25.1s) 17:38:34 ~ 17:38:59 (22.1s)

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) 1,615 ms 3,004 ms 3,239 ms 3,241 ms 1,874 ms

首文本(SSE delta) 7,934 ms 3,005 ms 3,240 ms 3,243 ms 1,876 ms

总耗时 8,218 ms 13,331 ms 3,240 ms 24,248 ms 20,827 ms

输入 10 10 9,543 10 10

输出 1,185 1,748 154 3,312 2,260

缓存创建

9,681

5m: 0 / 1h: 9,681

8,648

5m: 0 / 1h: 8,648

8,977

5m: 0 / 1h: 8,977

8,731

5m: 0 / 1h: 8,731

缓存读取 0 0 0 0 0

输出速度 burst 169.3 tok/s burst 157.7 tok/s 119.3 tok/s

响应

{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"UNKNOWN","nba_2024":"UNKNOWN","copa_america_2024":"UNKNOWN","euro2024":"UNKNOWN","nobel_peace_2024

```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

{"facts":{"super_bowl_lviii":"UNKNOWN","oscars2024_best_picture":"UNKNOWN","ucl_2024":"UNKNOWN","nba_2024":"UNKNOWN","copa_america_2024":"UNKNOWN","euro2024":"UNKNOWN","nobel_peace_2024":"UNKNOWN","wo

```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 17:36:02 ~ 17:36:17 (13.7s) 17:38:44 ~ 17:39:04 (19.1s) 17:34:37 ~ 17:34:47 (6.6s) 17:39:29 ~ 17:39:49 (18.0s) 17:40:27 ~ 17:40:57 (26.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=523593162611847847415835675964175439663871022518099755811829389932989763

HTTP 头(TTFB) 978 ms 1,602 ms 5,107 ms 2,565 ms 892 ms

首文本(SSE delta) 989 ms 1,603 ms 5,108 ms 2,569 ms 894 ms

总耗时 13,011 ms 18,014 ms 5,108 ms 17,267 ms 25,914 ms

输入 139 193 166 10 193

输出 2,223 2,225 14 1,864 3,120

缓存创建

5m: 0 / 1h: 0

5,981

5m: 0 / 1h: 5,981

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 184.9 tok/s 135.6 tok/s - 126.8 tok/s 124.7 tok/s

响应

```json
{"n": 72, "c": [2, 7, 10]}
```

```json
{"n":72,"c":[2,7,10]}
```

{"n":78,"c":[1,2,4]}

I'll count the digits in the string exactly.

The string is: `523593162611847847415835675964175439663871022518099755811829389932989763`

**Counting:**
- Length (n): 72 characters
- Count of 0s: 2 (pos

```json
{"n":72,"c":[2,7,10]}
```

总计

总输入(含缓存) 142,599 142,742 44,865 106,345 142,799

总输出 5,407 4,603 234 5,957 6,095

总缓存创建

43,198

5m: 0 / 1h: 43,198

42,387

5m: 0 / 1h: 42,387

38,155

5m: 0 / 1h: 38,155

42,446

5m: 0 / 1h: 42,446

总缓存读取 99,212 100,112 0 68,130 100,110

输出速度 184.6 tok/s 150.1 tok/s burst 140.7 tok/s 120.9 tok/s

总执行时间 48.5s 1m 1.8s 53.2s 1m 6.7s 1m 6.9s

总墙钟时间

11m 3.2s

含步骤间等待

13m 54.0s

含步骤间等待

9m 35.0s

含步骤间等待

14m 41.0s

含步骤间等待

15m 41.2s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0911

Anthropic: Claude Haiku 4.5

$0.0863

Anthropic: Claude Haiku 4.5

$0.0460

Anthropic: Claude Haiku 4.5

$0.0844

Anthropic: Claude Haiku 4.5

$0.0938

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%) direct (50%) direct (100%) direct (100%)

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

CDN - cloudflare - - -

ID 格式 msg_01VTgRDNYmCHH359X5RaMaFN msg_01Jp7t2zqjkEC9o9Ektid1MF 自定义 msg_01WxrhE3RuQPUMnCSP7cLfsz msg_01TxDHoMSLo67XkqSXV3ZUUn

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
O-Max	cf-ray: a12bb4760b73265a-LAX server: Photon-Edge set-cookie: _cfuvid=LVEnueVgaXPEmErdWaHGR5fBZg8iHBxQWYLZJ48X7Bg-1782639363.5223918-1.0.1.1-R1RL41ZaKDyankektLClpUW1TxNdSavj3fHvHS09cxw; HttpOnly; SameSite=None; Secure; Path=/; Domain=api.anthropic.com	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	1782639363422-oiqu6qe4bu0xw req_011CcVRxxvesaVkAct8J2Khx a12bb4760b73265a-LAX
O-Max	cf-ray: a12bb86b89baa55c-AMS server: cloudflare x-oneapi-request-id: 202606280938461626033718268d9d66Mao2p5L	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	a12bb86b89baa55c-AMS
M-	server: nginx	id_format: non_standard has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	body:id:non_standardbody:model:date_suffixbody:usage.cache_fields	—
R-kiro	via: 1.1 google cf-ray: 9a7955ce62a5d23bdb3fd1f95ff9ff91-IAD server: nginx	id_format: msg_prefix via_chain: ["1.1 google"] extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_9a7955ce62a5d23bdb3fd1f95ff9ff91 9a7955ce62a5d23bdb3fd1f95ff9ff91-IAD
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_011CcVSJXN4r5LPFvW31U2jp

协议指纹评分（机器自动）

总分 / 100 79 97 32 93 100

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 0.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 10.0 /10 ×15 0.0 /10 ×15 10.0 /10 ×15 10.0 /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 5.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

sliding 5m 10.0 /10 ×13 10.0 /10 ×13 5.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13

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

系统提示纯净 5.0 /10 ×8 10.0 /10 ×8 0.0 /10 ×8 0.0 /10 ×8 10.0 /10 ×8

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

知识截止 0.0 /10 ×7 10.0 /10 ×7 5.0 /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 0.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

inference_geo 10.0 /10 ×5 10.0 /10 ×5 0.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

延迟基线 10.0 /10 ×5 7.7 /10 ×5 2.4 /10 ×5 8.1 /10 ×5 10.0 /10 ×5

Req-ID 透传 10.0 /10 ×4 0.0 /10 ×4 0.0 /10 ×4 0.0 /10 ×4 10.0 /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 79 v3.25.0 97 v3.25.0 32 v3.25.0 93 v3.25.0 基准（参考）

雷达图例

共 21 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 档位思考量 tier_thinking_volume_match
21 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 6784 ms 8169 ms 7014 ms 10269 ms 10059 ms

输出速度 184.6 tok/s 150.1 tok/s burst 140.7 tok/s 120.9 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 ⓘ

$0.0911

$0.0863

$0.0460

$0.0844

$0.0938

缓存读取占比 ⓘ 70% 70% 0% 64% 70%

揭晓

通道

服务商 SSSAiCode ↗

通道O-Max

服务商 OmniaKey ↗

通道O-Max

服务商 AnyAIGC ↗

通道M-

服务商 ClaudeZZ ↗

通道R-kiro

官方基准

返回列表