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

2026/06/19 运行日期

claude-sonnet-4-6 官方基准

1 通道分数 < 70 另含 1 任务失败

已揭晓揭晓进度

步骤 / 指标

AIMZ

M-Max

SSSAiCode

O-Max

官方基准

SAIAi

O-Max

HongMaCC

u-cc

基本信息

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

模型 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6

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 03:51:44 ~ 03:51:49 (4.2s) 03:51:42 ~ 03:51:47 (3.1s) 03:51:48 ~ 03:51:53 (3.9s) 03:51:40 ~ 03:51:45 (3.9s) 03:51:46 ~ 03:51:51 (4.6s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,940 ms 1,892 ms 2,686 ms 2,750 ms 3,105 ms

首文本(SSE delta) 3,005 ms 2,003 ms 2,688 ms 2,751 ms 3,107 ms

总耗时 3,089 ms 2,154 ms 2,974 ms 2,909 ms 3,107 ms

输入 3 3 3 3 3

输出 5 5 5 5 5

缓存创建

31,985

5m: 0 / 1h: 31,985

31,988

5m: 0 / 1h: 31,988

5m: 0 / 1h: 0

32,318

5m: 0 / 1h: 32,318

28,384

5m: 0 / 1h: 28,384

缓存读取 0 0 32,318 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 03:55:14 ~ 03:55:19 (4.2s) 03:55:25 ~ 03:55:30 (3.0s) 03:54:46 ~ 03:54:51 (3.4s) 03:53:42 ~ 03:53:47 (3.4s) 03:55:49 ~ 03:55:54 (3.9s)

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) 2,594 ms 1,853 ms 1,931 ms 1,981 ms 2,822 ms

首文本(SSE delta) 2,913 ms 2,164 ms 1,932 ms 1,983 ms 2,823 ms

总耗时 2,972 ms 2,227 ms 2,373 ms 2,346 ms 2,823 ms

输入 3 3 3 3 3

输出 22 22 22 22 22

缓存创建

178

5m: 0 / 1h: 178

178

5m: 0 / 1h: 178

5m: 0 / 1h: 0

178

5m: 0 / 1h: 178

178

5m: 0 / 1h: 178

缓存读取 31,985 31,988 32,496 32,318 28,384

响应

vendor: anthropic
brand: claude
model: claude-sonnet-4-5

vendor: anthropic
brand: claude
model: claude-sonnet-4-5

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

步骤 3: cutoff text

时间 03:57:56 ~ 03:58:01 (4.5s) 03:56:33 ~ 03:56:38 (3.0s) 03:58:36 ~ 03:58:46 (3.8s) 03:57:19 ~ 03:57:29 (8.2s) 03:57:13 ~ 03:57:18 (4.2s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 2,996 ms 2,076 ms 2,717 ms 6,995 ms 3,105 ms

首文本(SSE delta) 3,160 ms 2,197 ms 2,811 ms 6,997 ms 3,106 ms

总耗时 3,315 ms 2,342 ms 2,898 ms 7,278 ms 3,106 ms

输入 3 3 3 3 3

输出 8 8 8 8 8

缓存创建

5m: 0 / 1h: 38

5m: 0 / 1h: 57

5m: 0 / 1h: 0

5m: 0 / 1h: 57

5m: 0 / 1h: 55

缓存读取 32,163 32,166 32,553 32,496 28,562

响应

2024-04

2025-04

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 04:01:46 ~ 04:01:51 (4.9s) 03:59:04 ~ 03:59:09 (3.4s) 04:02:12 ~ 04:02:23 (8.5s) 03:59:16 ~ 03:59:21 (3.8s) 04:00:18 ~ 04:00:23 (4.5s)

提示词

查看本步 Prompt

用中文说说你的身份：什么产品？由谁打造？50 字以内。

HTTP 头(TTFB) 2,717 ms 1,775 ms 6,562 ms 1,864 ms 3,012 ms

首文本(SSE delta) 3,700 ms 2,559 ms 6,578 ms 1,866 ms 3,013 ms

总耗时 3,794 ms 2,668 ms 7,505 ms 2,819 ms 3,013 ms

输入 3 3 3 3 3

输出 69 51 55 50 62

缓存创建

5m: 0 / 1h: 39

5m: 0 / 1h: 43

5m: 0 / 1h: 0

5m: 0 / 1h: 35

5m: 0 / 1h: 39

缓存读取 32,201 32,223 32,588 32,553 28,617

输出速度 - - 59.3 tok/s 52.5 tok/s burst

响应

我是 Claude，由 Anthropic 公司打造的 AI 助手。Anthropic 是一家专注于 AI 安全研究的公司，致力于开发安全、可靠且有益的人工智能系统。

我是 Claude，由 Anthropic 开发的 AI 助手。我擅长回答问题、分析代码、撰写文本和协助各类任务。

我是 Claude，由 Anthropic 开发的 AI 助手。我基于大型语言模型构建，旨在提供安全、有益的对话和任务协助。

我是Claude，由Anthropic开发的AI助手。我基于大型语言模型构建，旨在提供安全、有益的对话与任务协助。

我是 Claude，由 Anthropic 公司打造的 AI 助手。目前以 Claude Code 的形式运行，作为官方命令行工具帮助用户完成软件工程任务。

步骤 5: knowledge_recall text

时间 04:03:35 ~ 04:03:45 (5.6s) 04:02:55 ~ 04:03:00 (4.5s) 04:03:24 ~ 04:03:29 (4.6s) 04:01:32 ~ 04:01:42 (5.6s) 04:01:24 ~ 04:01:34 (5.9s)

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) 2,281 ms 1,420 ms 1,639 ms 1,638 ms 4,797 ms

首文本(SSE delta) 4,434 ms 3,640 ms 1,640 ms 1,639 ms 4,798 ms

总耗时 4,514 ms 3,704 ms 3,532 ms 4,641 ms 4,798 ms

输入 3 3 3 3 3

输出 198 198 205 205 198

缓存创建

7,698

5m: 0 / 1h: 7,698

7,706

5m: 0 / 1h: 7,706

2,314

5m: 0 / 1h: 2,314

8,027

5m: 0 / 1h: 8,027

15,649

5m: 0 / 1h: 15,649

缓存读取 0 0 5,718 0 0

输出速度 - - 108.4 tok/s 68.3 tok/s burst

响应

{"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","nobel

{"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","nobel

{"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","nobel

{"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","nobel

{"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","nobel

步骤 6: digit_count text

时间 04:05:02 ~ 04:05:17 (11.0s) 04:04:48 ~ 04:04:53 (0.9s) 04:04:50 ~ 04:05:00 (9.5s) 04:03:13 ~ 04:03:28 (14.2s) 04:04:10 ~ 04:04:40 (25.1s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=830412143350712882282155550005051881192612303568716643957608100849363449

HTTP 头(TTFB) 1,621 ms - 1,070 ms 1,342 ms 11,331 ms

首文本(SSE delta) 1,622 ms - 1,072 ms 1,343 ms 11,333 ms

总耗时 9,821 ms - 8,590 ms 13,230 ms 11,333 ms

输入 110 - 164 164 1

输出 630 - 627 1,097 693

缓存创建

5m: 0 / 1h: 0

199

5m: 0 / 1h: 199

缓存读取 0 - 0 0 13,463

输出速度 76.8 tok/s - 83.4 tok/s 92.3 tok/s burst

响应

```json
{"n":72,"c":[10,3,4]}
```

失败

评测失败

```json
{"n":72,"c":[10,3,4]}
```

```json
{"n":72,"c":[10,3,4]}
```

**Verification walk-through**

String (split for clarity):
```
8 3 0 4 1 2 1 4 3 3  ← pos 1-10
5 0 7 1 2 8 8 2 2 8  ← pos 11-20
2 1 5 5 5 5 0 0 0 5  ← pos 21-30
0 5

I'll count manually and carefully.

**s = `830412143350712882282155550005051881192612303568716643957608100849363449`**

Let me count the length and each digit:

**Length (n):**
```
8304121433507128822

总计

总输入(含缓存)

136,412

136,364

仅前 6 步

138,166

138,161

143,546

总输出

932

284

仅前 6 步

922

1,387

988

总缓存创建

39,938

5m: 0 / 1h: 39,938

39,972

5m: 0 / 1h: 39,972

仅前 6 步

2,314

5m: 0 / 1h: 2,314

40,615

5m: 0 / 1h: 40,615

44,504

5m: 0 / 1h: 44,504

总缓存读取

96,349

96,377

仅前 6 步

135,673

97,367

99,026

输出速度

76.8 tok/s

仅前 6 步

85.8 tok/s

85.3 tok/s

burst

总执行时间

34.4s

18.0s

仅前 6 步

33.7s

39.1s

48.2s

总墙钟时间

13m 32.9s

含步骤间等待

13m 10.9s

含步骤间等待

仅前 6 步

13m 12.0s

含步骤间等待

11m 47.7s

含步骤间等待

12m 54.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1930

Anthropic: Claude Sonnet 4.6

$0.1831

Anthropic: Claude Sonnet 4.6 (部分)

$0.0637

Anthropic: Claude Sonnet 4.6

$0.2029

Anthropic: Claude Sonnet 4.6

$0.2115

Anthropic: Claude Sonnet 4.6

通道指纹

来源步骤 digit_count knowledge_recall digit_count digit_count digit_count

平台 one-api (100%) direct (100%) direct (100%) direct (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_011JTq6bZf4yzjBPXHbJJjiY msg_01XofHAjmrVCCoSQvJc1sead msg_01KjiayPhFNHaxanePRTsuGY msg_01TeQiHq4gHPhdc5NZrDvL4S msg_01CNHScycGno2gkcE12y4hYd

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
M-Max	cf-ray: a0dce8192916c276-AMS server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606182005035988476698268d9d6tkZ5fTjw	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: false 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:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a0dce8192916c276-AMS
O-Max	cf-ray: a0dce4fc98ab1360-LAX server: Photon-Edge set-cookie: _cfuvid=JN74C77bL9VBn3y.cU4gWbiytZ4.tDNHzBnDjyLiIes-1781812976.0951502-1.0.1.1-1oDCjKGknHgHJEI.Om_lepgtkKul63AVKL5fNTIvS7k; 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: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	1781812976037-6dogcm9m2unl2 req_011CcBKgPodZPwznxYkpKwcF a0dce4fc98ab1360-LAX
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: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcBKptnNwbhNFsLw5fv9s
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: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcBKhiw6p46Gewf8ANXBU
u-cc	server: nginx x-new-api-version: 1.0.14 x-oneapi-request-id: 202606182004241006584548268d9d6XOk8q1Ak	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: false usage_has_cache_fields: true	hdr:x-new-api-versionhdr:x-oneapi-request-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—

协议指纹评分（机器自动）

总分 / 100 83 100 100 89

基线状态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

缓存 TTL 一致 10.0 /10 ×15 — 10.0 /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

缓存连续性 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

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

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

身份(结构) 6.7 /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 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

service_tier 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

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

流式投递 0.0 /10 ×5 — 10.0 /10 ×5 10.0 /10 ×5 0.0 /10 ×5

Req-ID 透传 0.0 /10 ×4 — 10.0 /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

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

综合结论

与基线相似度 83 v3.24.0 执行失败，不评分基准（参考） 100 v3.24.0 89 v3.24.0

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 流式投递 buffer_dump_match
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 缓存连续 cache_continuity_intra
13 知识召回 knowledge_recall_match
14 延迟基线 latency_baseline_match
15 原生 msg-ID anthropic_msg_id_format
16 Sliding 缓存 cache_sliding_correctness
17 结构化身份 identity_structured_match
18 信封自报一致 envelope_self_report_match
19 中文遵循 instruction_following_lang
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 4584 ms 2619 ms 4645 ms 5537 ms 4697 ms

输出速度 76.8 tok/s - 85.8 tok/s 85.3 tok/s burst

完成步骤 6/6 成功 5/6 成功 · 1 失败 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 ⓘ

$0.1930

$0.1831 (部分)

$0.0637

$0.2029

$0.2115

缓存读取占比 ⓘ 71% 71% (部分) 98% 70% 69%

揭晓

通道

服务商 AIMZ ↗

通道M-Max

服务商 SSSAiCode ↗

通道O-Max

官方基准

服务商 SAIAi ↗

通道O-Max

服务商 HongMaCC ↗

通道u-cc

返回列表