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

2026/06/26 运行日期

claude-sonnet-4-6 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

TopRouterCN

O-Max

官方基准

luminai

M-Kiro/Antigravity

AIMZ

M-Max

LinkAPI

O-Max

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 08:07:37 ~ 08:07:42 (4.3s) 08:07:39 ~ 08:07:44 (4.8s) 08:07:31 ~ 08:07:41 (5.8s) 08:07:35 ~ 08:07:40 (4.8s) 08:07:33 ~ 08:07:53 (19.9s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,390 ms 3,105 ms 4,583 ms 2,846 ms 18,229 ms

首文本(SSE delta) 3,391 ms 3,106 ms 4,601 ms 2,898 ms 18,258 ms

总耗时 3,466 ms 3,187 ms 4,647 ms 3,057 ms 18,446 ms

输入 1,467 3 2,309 3 3

输出 5 5 1 5 5

缓存创建

30,854

5m: 0 / 1h: 30,854

32,318

5m: 0 / 1h: 32,318

1,327

5m: 1,327 / 1h: 0

31,985

5m: 0 / 1h: 31,985

32,318

5m: 0 / 1h: 32,318

缓存读取 0 0 25,216 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 08:08:42 ~ 08:08:52 (5.6s) 08:08:51 ~ 08:08:56 (3.4s) 08:11:20 ~ 08:11:30 (5.9s) 08:10:43 ~ 08:10:53 (5.0s) 08:11:41 ~ 08:11:52 (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) 4,791 ms 1,935 ms 4,485 ms 2,556 ms 4,720 ms

首文本(SSE delta) 4,792 ms 1,937 ms 4,486 ms 2,869 ms 4,721 ms

总耗时 4,864 ms 2,418 ms 4,674 ms 2,963 ms 5,212 ms

输入 3 3 2,322 3 3

输出 22 22 24 22 22

缓存创建

1,642

5m: 0 / 1h: 1,642

178

5m: 0 / 1h: 178

1,335

5m: 1,335 / 1h: 0

178

5m: 0 / 1h: 178

178

5m: 0 / 1h: 178

缓存读取 30,854 32,318 25,366 31,985 32,318

响应

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

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

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

步骤 3: cutoff text

时间 08:11:49 ~ 08:11:59 (5.9s) 08:12:44 ~ 08:12:54 (6.1s) 08:15:20 ~ 08:15:30 (6.8s) 08:12:11 ~ 08:12:16 (4.3s) 08:14:06 ~ 08:14:16 (6.2s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 5,034 ms 4,436 ms 4,227 ms 3,057 ms 4,619 ms

首文本(SSE delta) 5,035 ms 4,608 ms 4,371 ms 3,112 ms 4,744 ms

总耗时 5,119 ms 4,771 ms 4,423 ms 3,265 ms 4,872 ms

输入 3 3 2,326 3 3

输出 8 8 4 8 8

缓存创建

5m: 0 / 1h: 38

5m: 0 / 1h: 55

1,336

5m: 1,336 / 1h: 0

5m: 0 / 1h: 38

5m: 0 / 1h: 57

缓存读取 32,496 32,496 25,402 32,163 32,496

响应

2025-08

2025-08

2025-08

2025-04

2025-08

步骤 4: identity_free text

时间 08:13:35 ~ 08:13:45 (5.2s) 08:15:02 ~ 08:15:07 (4.6s) 08:18:51 ~ 08:19:01 (6.2s) 08:14:35 ~ 08:14:45 (5.9s) 08:16:02 ~ 08:16:07 (4.5s)

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) 4,144 ms 2,610 ms 4,191 ms 3,351 ms 2,878 ms

首文本(SSE delta) 4,145 ms 2,612 ms 4,815 ms 3,668 ms 3,332 ms

总耗时 4,223 ms 3,251 ms 4,864 ms 3,813 ms 3,521 ms

输入 3 3 2,329 3 3

输出 49 50 49 26 41

缓存创建

5m: 0 / 1h: 43

1,339

5m: 1,339 / 1h: 0

5m: 0 / 1h: 39

5m: 0 / 1h: 41

缓存读取 32,534 32,551 25,443 32,201 32,553

输出速度 - 78.2 tok/s burst - -

响应

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助编写代码、分析问题、回答问题等多种任务。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助回答问题、分析代码、撰写文本等多种任务。

我是 Claude，由 Anthropic 开发的 AI 助手。当前版本为 Claude Sonnet 4.6，擅长编程、写作、分析等多种任务。

我是 Claude，由 Anthropic 公司打造的 AI 助手。

我叫 Claude，是由 Anthropic 开发的 AI 助手。当前版本为 Claude Sonnet 4.6。

步骤 5: knowledge_recall text

时间 08:16:42 ~ 08:16:52 (5.6s) 08:17:36 ~ 08:17:41 (4.6s) 08:20:19 ~ 08:20:29 (6.7s) 08:16:31 ~ 08:16:46 (10.1s) 08:18:23 ~ 08:18:33 (6.9s)

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) 4,790 ms 1,692 ms 3,314 ms 4,099 ms 3,680 ms

首文本(SSE delta) 4,791 ms 1,694 ms 5,278 ms 6,179 ms 3,698 ms

总耗时 4,841 ms 3,564 ms 5,439 ms 6,261 ms 5,927 ms

输入 3 3 597 3 3

输出 204 204 211 198 204

缓存创建

8,015

5m: 0 / 1h: 8,015

8,033

5m: 0 / 1h: 8,033

343

5m: 343 / 1h: 0

7,655

5m: 0 / 1h: 7,655

8,024

5m: 0 / 1h: 8,024

缓存读取 0 0 6,520 0 0

输出速度 burst 109.1 tok/s burst - 91.5 tok/s

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 08:18:00 ~ 08:18:15 (14.7s) 08:20:50 ~ 08:21:05 (10.8s) 08:23:06 ~ 08:23:11 (4.9s) 08:19:18 ~ 08:19:38 (16.9s) 08:21:30 ~ 08:22:05 (33.3s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=954670213212827717871503028997558581986426057104115631258340361633689754

HTTP 头(TTFB) 6,052 ms 1,463 ms 3,451 ms 1,381 ms 16,388 ms

首文本(SSE delta) 6,053 ms 1,464 ms 3,742 ms 1,780 ms 16,389 ms

总耗时 13,854 ms 9,809 ms 3,796 ms 15,862 ms 32,384 ms

输入 164 164 16 110 164

输出 1,193 805 14 1,296 1,644

缓存创建

5m: 0 / 1h: 0

5m: 8 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 165 0 0

输出速度 burst 96.5 tok/s - 92.0 tok/s 102.8 tok/s

响应

## Counting digits in `s`

**s** = `954670213212827717871503028997558581986426057104115631258340361633689754`

### Step 1 — Length

Grouping in tens:
`9546702132` `1282771787` `1503028997` `5585819864

```json
{"n":72,"c":[6,8,5]}
```

**Verification walkthrough:**

String: `954670213212827717871503028997558581986426057104115631258340361633689754`

| Digit | Positions (1-indexed) | Count |
|---|---|

{"n":71,"c":[3,7,5]}

```json
{"n":72,"c":[6,8,5]}
```

**Verification breakdown** for `s = 954670213212827717871503028997558581986426057104115631258340361633689754`:

- **Length**: 72 digits
- **0s** (count=6): positions

## Counting digits in s

**s** = `954670213212827717871503028997558581986426057104115631258340361633689754`

### Step 1 — Length
Splitting into groups of 10:
```
9546702132  (10)
1282771787  (10)
1503

总计

总输入(含缓存) 138,119 138,171 123,699 136,369 138,164

总输出 1,481 1,094 303 1,555 1,924

总缓存创建

40,592

5m: 0 / 1h: 40,592

40,627

5m: 0 / 1h: 40,627

5,688

5m: 5,688 / 1h: 0

39,895

5m: 0 / 1h: 39,895

40,618

5m: 0 / 1h: 40,618

总缓存读取 95,884 97,365 108,112 96,349 97,367

输出速度 burst 97.6 tok/s burst 92.0 tok/s 101.4 tok/s

总执行时间 41.1s 34.3s 36.3s 46.9s 1m 17.1s

总墙钟时间

10m 38.0s

含步骤间等待

13m 26.0s

含步骤间等待

15m 40.0s

含步骤间等待

12m 3.0s

含步骤间等待

14m 32.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2081

Anthropic: Claude Sonnet 4.6

$0.1985

Anthropic: Claude Sonnet 4.6

$0.0800

Anthropic: Claude Sonnet 4.6

$0.2022

Anthropic: Claude Sonnet 4.6

$0.2109

Anthropic: Claude Sonnet 4.6

通道指纹

来源步骤 digit_count digit_count digit_count digit_count digit_count

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

ID 格式 msg_01UPTVK8TDAvhE18pMGxJEGG msg_01NW4psiLcrWqpuXF6bFcHxT msg_01AvccbVsLvWxHSSVkGwHAki9r msg_012EKSvJ9bAdyCxWGEQkWnF9 msg_01JL3pG7mCXpBDERk3tnYUSV

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/1.27.5	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	tr-ew7y4grd-k7ftlc0abs-001446
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_011CcQv1cAp2twVvU6CFeqD7
M-Kiro/Antigravity	cf-ray: a1180fc25d67f5e7-AMS server: cloudflare x-oneapi-request-id: 202606260023083758386138268d9d6ceNhim4Y	id_format: msg_prefix extra_fields: ["container"] has_service_tier: true has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	hdr:x-oneapi-request-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a1180fc25d67f5e7-AMS
M-Max	cf-ray: a1180a307a6c1716-AMS server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606260019197931868318268d9d6aa85VJrf	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	a1180a307a6c1716-AMS
O-Max	server: nginx set-cookie: server_name_session=7d88fca8b1426ca76c58d5aeba6fb224; Max-Age=86400; httponly; path=/ x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606260021316666820508268d9d6Aa2PEn4Z	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 90 96 73 83 93

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 6.9 /10 ×20 7.3 /10 ×20 10.0 /10 ×20 7.3 /10 ×20 7.3 /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 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 10.0 /10 ×8

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

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

知识截止 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 5.0 /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 0.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

延迟基线 9.5 /10 ×5 10.0 /10 ×5 8.8 /10 ×5 10.0 /10 ×5 7.1 /10 ×5

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

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

综合结论

与基线相似度 90 v3.24.1 基准（参考） 73 v3.24.1 83 v3.24.1 93 v3.24.1

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 6061 ms 4500 ms 4641 ms 5870 ms 11727 ms

输出速度 burst 97.6 tok/s burst 92.0 tok/s 101.4 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 ⓘ

$0.2081

$0.1985

$0.0800

$0.2022

$0.2109

缓存读取占比 ⓘ 69% 70% 87% 71% 70%

揭晓

通道

服务商 TopRouterCN ↗

通道O-Max

官方基准

服务商 luminai ↗

通道M-Kiro/Antigravity

服务商 AIMZ ↗

通道M-Max

服务商 LinkAPI ↗

通道O-Max

返回列表