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

2026/07/01 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

InfAI

R-Kiro

SSSAiCode

O-Max

LingxiCode

M-

LinkAPI

O-Max

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 14:24:50 ~ 14:25:00 (5.0s) 14:24:48 ~ 14:24:59 (7.9s) 14:24:46 ~ 14:24:56 (5.8s) 14:24:42 ~ 14:24:52 (7.4s) 14:24:44 ~ 14:24:54 (5.2s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,146 ms 2,707 ms 1,764 ms 3,296 ms 1,839 ms

首文本(SSE delta) 2,147 ms 2,708 ms 1,994 ms 3,314 ms 1,843 ms

总耗时 2,726 ms 2,792 ms 2,141 ms 3,822 ms 3,445 ms

输入 10 4 10 10 10

输出 56 1 43 55 58

缓存创建

6,166

5m: 0 / 1h: 6,166

20,626

5m: 0 / 1h: 20,626

21,180

5m: 0 / 1h: 21,180

6,290

5m: 0 / 1h: 6,167

21,545

5m: 0 / 1h: 21,545

缓存读取 15,379 0 0 15,686 0

输出速度 96.7 tok/s - - 108.3 tok/s 36.2 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 14:27:09 ~ 14:27:19 (5.1s) 14:26:32 ~ 14:26:37 (4.3s) 14:27:17 ~ 14:27:22 (4.6s) 14:27:19 ~ 14:27:29 (7.0s) 14:26:16 ~ 14:26:26 (6.3s)

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。请按下列 3 行格式精确填空作答（label 不变，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) 1,757 ms 2,656 ms 1,883 ms 3,544 ms 1,882 ms

首文本(SSE delta) 1,758 ms 2,657 ms 3,684 ms 3,545 ms 1,883 ms

总耗时 4,054 ms 2,857 ms 3,843 ms 5,797 ms 4,270 ms

输入 10 12 10 10 10

输出 199 16 245 239 212

缓存创建

240

5m: 0 / 1h: 240

122

5m: 0 / 1h: 122

227

5m: 0 / 1h: 227

243

5m: 0 / 1h: 239

242

5m: 0 / 1h: 242

缓存读取 21,545 20,626 21,180 21,976 21,545

输出速度 86.7 tok/s - burst 106.1 tok/s 88.8 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-20241022

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 14:30:18 ~ 14:30:23 (3.9s) 14:27:56 ~ 14:28:06 (5.7s) 14:29:53 ~ 14:29:58 (3.8s) 14:28:36 ~ 14:28:42 (4.0s) 14:27:39 ~ 14:27:44 (4.0s)

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) 1,819 ms 3,167 ms 2,503 ms 2,558 ms 1,911 ms

首文本(SSE delta) 1,821 ms 3,168 ms 2,848 ms 2,559 ms 1,955 ms

总耗时 2,845 ms 4,267 ms 3,004 ms 3,145 ms 2,646 ms

输入 10 20 10 10 10

输出 100 2 54 61 66

缓存创建

239

5m: 0 / 1h: 239

5m: 0 / 1h: 36

268

5m: 0 / 1h: 268

267

5m: 0 / 1h: 262

252

5m: 0 / 1h: 252

缓存读取 21,785 20,748 21,407 22,220 21,787

输出速度 97.7 tok/s - - 104.1 tok/s 95.5 tok/s

响应

2025-02

2025-02

2024-04

2025-02

2025-02

步骤 4: identity_free text

时间 14:33:22 ~ 14:33:32 (5.2s) 14:30:33 ~ 14:30:43 (8.7s) 14:33:44 ~ 14:33:54 (5.2s) 14:31:38 ~ 14:31:48 (9.4s) 14:31:30 ~ 14:31:40 (5.2s)

提示词

查看本步 Prompt

用中文回答：你是谁？由谁开发？50 字以内。

HTTP 头(TTFB) 2,321 ms 4,891 ms 1,873 ms 2,645 ms 2,073 ms

首文本(SSE delta) 2,322 ms 4,892 ms 4,232 ms 2,646 ms 2,074 ms

总耗时 4,102 ms 5,636 ms 4,387 ms 8,693 ms 4,058 ms

输入 10 28 10 10 10

输出 213 43 370 943 208

缓存创建

134

5m: 0 / 1h: 134

20,801

5m: 0 / 1h: 20,801

5m: 0 / 1h: 96

100

5m: 0 / 1h: 99

106

5m: 0 / 1h: 106

缓存读取 22,024 0 21,675 22,487 22,039

输出速度 119.7 tok/s 57.8 tok/s burst 155.9 tok/s 104.8 tok/s

响应

我是Claude，由Anthropic公司开发的AI助手。我可以帮助用户完成各种软件工程任务，包括代码编写、调试、分析和解释。

我是 Kiro，由 Anthropic 开发的 AI 助手。我基于 Claude 大语言模型构建，帮助用户完成软件工程任务，包括编码、调试、重构等工作。

我是Claude，由Anthropic开发的AI助手。我可以帮助编码、分析、写作等各类任务，致力于提供有帮助且诚实的回应。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我能帮助用户完成代码编写、分析、写作等各类任务。

我是Claude，由Anthropic公司开发的AI助手。属于Claude产品系列，可通过Claude API、网页版Claude.ai、Claude Code等多个界面使用。

步骤 5: knowledge_recall text

时间 14:34:32 ~ 14:34:57 (22.7s) 14:32:31 ~ 14:32:41 (5.1s) 14:36:09 ~ 14:36:24 (11.3s) 14:33:05 ~ 14:33:25 (18.7s) 14:32:46 ~ 14:33:11 (20.8s)

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) 1,398 ms 2,333 ms 1,597 ms 2,171 ms 1,761 ms

首文本(SSE delta) 1,400 ms 2,335 ms 9,805 ms 2,172 ms 1,789 ms

总耗时 21,724 ms 3,232 ms 10,495 ms 17,538 ms 19,674 ms

输入 10 36 10 10 10

输出 2,405 136 1,519 2,731 2,045

缓存创建

9,322

5m: 0 / 1h: 9,322

9,059

5m: 0 / 1h: 9,059

9,092

5m: 0 / 1h: 9,092

10,240

5m: 0 / 1h: 10,040

9,304

5m: 0 / 1h: 9,304

缓存读取 0 0 0 0 0

输出速度 118.3 tok/s 151.6 tok/s - 177.7 tok/s 114.3 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_20

```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 14:38:35 ~ 14:39:00 (24.3s) 14:34:34 ~ 14:34:44 (5.9s) 14:38:02 ~ 14:38:27 (23.1s) 14:36:07 ~ 14:36:27 (15.1s) 14:36:23 ~ 14:36:53 (25.2s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=734250423092597290770578692367625177511546126231626671483746866602451099

HTTP 头(TTFB) 1,842 ms 4,467 ms 1,732 ms 1,737 ms 1,777 ms

首文本(SSE delta) 1,844 ms 4,468 ms 1,734 ms 1,738 ms 1,778 ms

总耗时 23,132 ms 4,468 ms 22,442 ms 13,960 ms 24,197 ms

输入 193 145 139 196 2,034

输出 3,371 9 4,397 1,979 2,743

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 158.4 tok/s - 212.3 tok/s 161.9 tok/s 122.4 tok/s

响应

```json
{"n": 72, "c": [6, 10, 6]}
```

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

```json
{"n": 72, "c": [6, 10, 6]}
```

```json
{"n":72,"c":[6,10,6]}
```

```json
{"n":72,"c":[6,10,6]}
```

总计

总输入(含缓存) 97,077 92,263 95,314 99,755 98,904

总输出 6,344 207 6,628 6,008 5,332

总缓存创建

16,101

5m: 0 / 1h: 16,101

50,644

5m: 0 / 1h: 50,644

30,863

5m: 0 / 1h: 30,863

17,140

5m: 0 / 1h: 16,807

31,449

5m: 0 / 1h: 31,449

总缓存读取 80,733 41,374 64,262 82,369 65,371

输出速度 134.1 tok/s 109.1 tok/s 212.3 tok/s 162.5 tok/s 113.5 tok/s

总执行时间 1m 6.1s 37.6s 53.9s 1m 1.6s 1m 6.7s

总墙钟时间

14m 10.0s

含步骤间等待

9m 56.0s

含步骤间等待

13m 41.1s

含步骤间等待

11m 45.0s

含步骤间等待

12m 9.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0602

Anthropic: Claude Haiku 4.5

$0.0687

Anthropic: Claude Haiku 4.5

$0.0783

Anthropic: Claude Haiku 4.5

$0.0595

Anthropic: Claude Haiku 4.5

$0.0746

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

ID 格式 msg_01U4CVsD3F8DaYeU7F2etqHx msg_aa5af98b6baa4b2c834c9616a8c80d27 msg_01XXqfBF9cSBifeaQvDpNQGg msg_01BVcws4hFryZPkrJmqFXHTL msg_01F5P5pZGKCDCa7exXANi8Wj

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_011Ccasrv3PzcDe5jdYh8CQ2
R-Kiro	server: nginx	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	body:id:msg_prefixbody:model:date_suffixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	20260701063437244492728268d9d6GA1g7NH3
O-Max	cf-ray: a14367d8ed9ac28f-LAX server: Photon-Edge	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	1782887883571-dce4zhpavw7wn req_011CcaspTQhy4BV85UoURP1V a14367d8ed9ac28f-LAX
M-	cf-ray: a143650aab81734b-NRT server: cloudflare set-cookie: server_name_session=f536c2dc7af91930135bd6223ed2791c; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 20260701063609374426155eFTmzC9B	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	a143650aab81734b-NRT
O-Max	server: nginx set-cookie: server_name_session=c576e09cb6f84c78844c48bf7e3b48cb; Max-Age=86400; httponly; path=/ x-new-api-version: v1.0.0-rc.15 x-oneapi-request-id: 202607010636247224819568268d9d6aYq7AJah	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-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—

协议指纹评分（机器自动）

总分 / 100 100 55 79 97 97

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 0.1 /10 ×20 8.6 /10 ×20 10.0 /10 ×20 8.5 /10 ×20

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

延迟基线 10.0 /10 ×15 9.6 /10 ×15 10.0 /10 ×15 9.9 /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 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

sliding 5m 10.0 /10 ×13 7.7 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13

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

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

身份(自由) 10.0 /10 ×7 0.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7

service_tier 10.0 /10 ×6 0.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

知识截止 10.0 /10 ×5 10.0 /10 ×5 0.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

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

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

Req-ID 透传 10.0 /10 ×2 0.0 /10 ×2 10.0 /10 ×2 0.0 /10 ×2 0.0 /10 ×2

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） 55 v3.26.0 79 v3.26.0 97 v3.26.0 97 v3.26.0

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 9764 ms 3875 ms 7719 ms 8826 ms 9715 ms

输出速度 134.1 tok/s 109.1 tok/s 212.3 tok/s 162.5 tok/s 113.5 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 ⓘ

$0.0602

$0.0687

$0.0783

$0.0595

$0.0746

缓存读取占比 ⓘ 83% 45% 67% 83% 66%

揭晓

通道

官方基准

服务商 InfAI ↗

通道R-Kiro

服务商 SSSAiCode ↗

通道O-Max

服务商 LingxiCode ↗

通道M-

服务商 LinkAPI ↗

通道O-Max

返回列表