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

2026/06/23 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

SAIAi

O-Max

SSSAiCode

R-Kiro

官方基准

luminai

M-Kiro/Antigravity

ClaudeCN

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 v2.1.160 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 11:12:17 ~ 11:12:22 (3.8s) 11:12:15 ~ 11:12:25 (5.0s) 11:12:19 ~ 11:12:24 (3.4s) 11:12:11 ~ 11:12:21 (6.4s) 11:12:13 ~ 11:12:23 (6.2s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,223 ms 2,447 ms 1,587 ms 4,734 ms 3,737 ms

首文本(SSE delta) 2,225 ms 2,449 ms 1,588 ms 4,775 ms 3,739 ms

总耗时 2,671 ms 2,980 ms 2,002 ms 4,850 ms 4,448 ms

输入 10 0 10 1,443 10

输出 53 65 60 1 52

缓存创建

33,122

5m: 0 / 1h: 33,122

5m: 0 / 1h: 0

4,111

5m: 4,111 / 1h: 0

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 33,122 23,301 0

输出速度 118.8 tok/s 122.4 tok/s 144.9 tok/s - 73.3 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 11:14:18 ~ 11:14:23 (4.7s) 11:14:16 ~ 11:14:26 (5.4s) 11:16:20 ~ 11:16:25 (4.7s) 11:16:07 ~ 11:16:32 (24.5s) 11:14:12 ~ 11:14:22 (5.8s)

提示词

查看本步 Prompt

输出 3 行 metadata 描述你自己（当前 API 模型）。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。禁止任何额外字符：
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) 1,583 ms 1,813 ms 1,638 ms 23,260 ms 4,108 ms

首文本(SSE delta) 1,606 ms 1,844 ms 1,640 ms 23,261 ms 4,109 ms

总耗时 3,718 ms 3,935 ms 3,365 ms 23,375 ms 4,769 ms

输入 10 0 10 1,452 10

输出 262 326 199 28 208

缓存创建

244

5m: 0 / 1h: 244

251

5m: 0 / 1h: 251

4,137

5m: 4,137 / 1h: 0

243

5m: 0 / 1h: 243

缓存读取 33,122 0 33,122 23,445 33,122

输出速度 124.1 tok/s 155.9 tok/s 115.4 tok/s - burst

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-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 11:17:45 ~ 11:17:50 (3.6s) 11:15:31 ~ 11:15:36 (3.0s) 11:19:57 ~ 11:20:02 (3.1s) 11:18:24 ~ 11:18:34 (7.4s) 11:16:09 ~ 11:16:19 (6.0s)

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) 1,667 ms 1,697 ms 1,294 ms 4,653 ms 3,712 ms

首文本(SSE delta) 1,669 ms 1,698 ms 1,295 ms 4,689 ms 3,713 ms

总耗时 2,299 ms 2,269 ms 1,814 ms 4,970 ms 4,655 ms

输入 10 0 10 1,456 10

输出 75 79 68 4 73

缓存创建

302

5m: 0 / 1h: 302

241

5m: 0 / 1h: 241

4,146

5m: 4,146 / 1h: 0

248

5m: 0 / 1h: 248

缓存读取 33,366 0 33,373 23,499 33,365

输出速度 119.0 tok/s 138.4 tok/s 131.0 tok/s - 77.5 tok/s

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 11:19:00 ~ 11:19:15 (13.6s) 11:18:17 ~ 11:18:27 (7.6s) 11:23:12 ~ 11:23:22 (8.0s) 11:20:03 ~ 11:20:28 (20.7s) 11:19:31 ~ 11:19:41 (8.9s)

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) 3,103 ms 2,504 ms 1,989 ms 19,453 ms 3,598 ms

首文本(SSE delta) 3,108 ms 2,506 ms 1,991 ms 19,556 ms 3,599 ms

总耗时 12,499 ms 6,768 ms 6,777 ms 19,634 ms 7,334 ms

输入 10 0 10 1,457 10

输出 1,254 699 726 55 456

缓存创建

109

5m: 0 / 1h: 109

110

5m: 0 / 1h: 110

4,151

5m: 4,151 / 1h: 0

115

5m: 0 / 1h: 115

缓存读取 33,668 0 33,614 23,527 33,613

输出速度 133.5 tok/s 164.0 tok/s 151.7 tok/s - 122.1 tok/s

响应

我是Claude，由Anthropic公司开发的AI助手。我可以帮助用户进行编程、分析、写作等多种任务。

我是Kiro，由Anthropic开发的AI助手，基于Claude模型，帮助开发者进行软件工程工作。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。可以帮助您完成软件工程任务，包括代码编写、调试、审查等工作。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我可以帮助您进行代码编写、问题分析、文本处理等各种任务。

我是Claude，由Anthropic开发的AI助手。能帮助编程、代码分析、问题解决等技术工作。

步骤 5: knowledge_recall text

时间 11:21:21 ~ 11:21:47 (20.9s) 11:19:48 ~ 11:20:08 (15.9s) 11:26:13 ~ 11:27:03 (49.6s) 11:22:49 ~ 11:23:29 (36.5s) 11:22:39 ~ 11:22:59 (18.6s)

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) 1,506 ms 1,886 ms 1,395 ms 32,752 ms 2,509 ms

首文本(SSE delta) 1,508 ms 1,888 ms 1,397 ms 32,753 ms 2,510 ms

总耗时 19,710 ms 14,897 ms 48,471 ms 35,074 ms 17,511 ms

输入 10 0 10 375 10

输出 2,809 2,337 6,662 201 2,241

缓存创建

9,649

5m: 0 / 1h: 9,649

3,347

5m: 0 / 1h: 3,347

1,067

5m: 1,067 / 1h: 0

8,802

5m: 0 / 1h: 8,802

缓存读取 0 0 5,721 6,050 0

输出速度 154.3 tok/s 179.6 tok/s 141.5 tok/s 86.6 tok/s 149.4 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":"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": "Real Madrid",
    "nba_2024": "Boston Celtics",
    "copa_america_20

步骤 6: digit_count text

时间 11:25:07 ~ 11:25:27 (19.0s) 11:24:00 ~ 11:24:10 (7.9s) 11:29:25 ~ 11:29:45 (17.4s) 11:26:03 ~ 11:26:43 (37.8s) 11:25:21 ~ 11:25:36 (13.0s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=542479110259084970510785891969498120905915571923930238581072634490510122

HTTP 头(TTFB) 1,012 ms 1,097 ms 1,161 ms 35,469 ms 1,699 ms

首文本(SSE delta) 1,014 ms 1,118 ms 1,162 ms 35,471 ms 1,700 ms

总耗时 17,885 ms 7,141 ms 16,111 ms 35,893 ms 11,510 ms

输入 193 0 193 190 193

输出 3,107 1,005 2,687 19 1,794

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 - 0

输出速度 184.2 tok/s 166.9 tok/s 179.7 tok/s - 182.9 tok/s

响应

```json
{"n":72,"c":[10,5,12]}
```

```json
{"n":72,"c":[10,3,5]}
```

```json
{"n":72,"c":[10,5,12]}
```

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

```json
{"n":72,"c":[10,5,12]}
```

总计

总输入(含缓存) 143,825 0 143,144 123,807 142,873

总输出 7,560 4,511 10,402 308 4,824

总缓存创建

43,426

5m: 0 / 1h: 43,426

3,949

5m: 0 / 1h: 3,949

17,612

5m: 17,612 / 1h: 0

42,530

5m: 0 / 1h: 42,530

总缓存读取 100,156 0 138,952 99,822 100,100

输出速度 158.7 tok/s 170.3 tok/s 149.7 tok/s 86.6 tok/s 152.9 tok/s

总执行时间 1m 5.5s 44.9s 1m 26.3s 2m 13.3s 58.5s

总墙钟时间

13m 10.0s

含步骤间等待

11m 55.2s

含步骤间等待

17m 26.1s

含步骤间等待

14m 32.2s

含步骤间等待

13m 23.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1023

Anthropic: Claude Haiku 4.5

>=$0.0226

Anthropic: Claude Haiku 4.5 (成本不完整)

$0.0711

Anthropic: Claude Haiku 4.5

$0.0317

Anthropic: Claude Haiku 4.5

$0.0875

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 (75%) direct (100%) one-api (100%) one-api (100%)

上游 Anthropic 直连 (100%) Anthropic 直连 (75%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%)

CDN - - - cloudflare cloudflare

ID 格式 msg_012Dbsd9H9sFwXoEbqrSnMMe msg_87b206296a034c45a8b1f4d0fe98ab77 msg_01EV1vHqRqamS71J51TGrZAV msg_016ZuJgqprdpJQpturRWp8ar3A msg_01YSEm4cd1q2Q8oZk9vGRGxC

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	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_011CcKUdsaimmhvKfT42sAtd
R-Kiro	server: Photon-Edge x-amzn-requestid: 80106e4a-1efc-4b16-8c7b-b3705b7b05e7	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.cache_fieldsupstream:x-amzn-requestidupstream:anthropic-direct-from-body	1782185041194-ut7mj6xeq8per 80106e4a-1efc-4b16-8c7b-b3705b7b05e7
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_011CcKUxv1bfFJrZQaagUMXH
M-Kiro/Antigravity	cf-ray: a10063a37d0b8f37-AMS server: cloudflare x-oneapi-request-id: 202606230326054577620128268d9d6d2OHzJ58	id_format: msg_prefix extra_fields: ["container"] has_service_tier: true has_inference_geo: false 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:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a10063a37d0b8f37-AMS
O-Max	cf-ray: a1006299cb3c0a7b-AMS server: cloudflare x-oneapi-request-id: 20260623112523391118626zyATL1XU	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	a1006299cb3c0a7b-AMS

协议指纹评分（机器自动）

总分 / 100 100 48 98 70 98

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

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

sliding 5m 10.0 /10 ×13 5.0 /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 未计入 10.0 /10 ×8 0.0 /10 ×8 10.0 /10 ×8

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

知识截止 10.0 /10 ×7 10.0 /10 ×7 10.0 /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

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

延迟基线 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 4.2 /10 ×5 10.0 /10 ×5

Req-ID 透传 10.0 /10 ×4 0.0 /10 ×4 10.0 /10 ×4 0.0 /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 100 v3.24.0 48 v3.24.0 基准（参考） 70 v3.24.0 98 v3.24.0

雷达图例

共 23 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 中文遵循 instruction_following_lang
22 档位思考量 tier_thinking_volume_match
23 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 9797 ms 6332 ms 13090 ms 20633 ms 8371 ms

输出速度 158.7 tok/s 170.3 tok/s 149.7 tok/s 86.6 tok/s 152.9 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 ⓘ

$0.1023

>=$0.0226 (成本不完整)

$0.0711

$0.0317

$0.0875

缓存读取占比 ⓘ 70% - 97% 81% 70%

揭晓

通道

服务商 SAIAi ↗

通道O-Max

服务商 SSSAiCode ↗

通道R-Kiro

官方基准

服务商 luminai ↗

通道M-Kiro/Antigravity

服务商 ClaudeCN ↗

通道O-Max

返回列表