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

2026/06/29 运行日期

claude-sonnet-4-6 官方基准

1 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

SudoCode

M-Max

官方基准

ClaudeCN

O-Max

AiTokensFlux

O-Pro

ClaudeAPI

O-Bdrk

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 01:31:05 ~ 01:31:10 (4.6s) 01:31:11 ~ 01:31:16 (4.2s) 01:31:03 ~ 01:31:13 (5.5s) 01:31:09 ~ 01:31:14 (3.9s) 01:31:07 ~ 01:31:17 (5.2s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,170 ms 3,047 ms 3,836 ms 2,872 ms 2,815 ms

首文本(SSE delta) 3,171 ms 3,049 ms 3,837 ms 2,873 ms 2,817 ms

总耗时 3,403 ms 3,155 ms 4,091 ms 3,033 ms 3,234 ms

输入 3 3 3 3 3

输出 5 5 5 5 5

缓存创建

32,318

5m: 0 / 1h: 32,318

32,318

5m: 0 / 1h: 32,318

32,319

5m: 0 / 1h: 32,319

32,457

5m: 0 / 1h: 32,457

32,318

5m: 32,318 / 1h: 0

缓存读取 0 0 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 01:33:29 ~ 01:33:39 (6.4s) 01:34:28 ~ 01:34:33 (3.3s) 01:34:26 ~ 01:34:36 (5.3s) 01:34:40 ~ 01:34:45 (3.5s) 01:34:11 ~ 01:34:16 (4.6s)

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) 4,058 ms 2,009 ms 3,551 ms 2,395 ms 1,927 ms

首文本(SSE delta) 4,077 ms 2,013 ms 3,552 ms 2,396 ms 1,928 ms

总耗时 5,433 ms 2,278 ms 3,974 ms 2,784 ms 2,809 ms

输入 3 3 3 3 3

输出 22 22 22 22 22

缓存创建

189

5m: 0 / 1h: 189

189

5m: 0 / 1h: 189

189

5m: 0 / 1h: 189

189

5m: 0 / 1h: 189

189

5m: 189 / 1h: 0

缓存读取 32,318 32,318 32,319 32,457 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-4-6

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

步骤 3: cutoff text

时间 01:35:42 ~ 01:35:47 (4.4s) 01:37:53 ~ 01:37:58 (4.1s) 01:36:15 ~ 01:36:25 (5.0s) 01:37:07 ~ 01:37:12 (4.5s) 01:37:10 ~ 01:37:20 (5.8s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 3,112 ms 2,358 ms 3,434 ms 3,472 ms 3,922 ms

首文本(SSE delta) 3,113 ms 2,360 ms 3,435 ms 3,473 ms 4,025 ms

总耗时 3,388 ms 2,839 ms 3,677 ms 3,706 ms 4,370 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: 55

5m: 0 / 1h: 57

5m: 0 / 1h: 55

32,562

5m: 32,562 / 1h: 0

缓存读取 32,507 32,507 32,508 32,646 0

响应

2025-08

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 01:39:16 ~ 01:39:26 (5.1s) 01:41:00 ~ 01:41:05 (4.3s) 01:39:49 ~ 01:39:59 (5.9s) 01:41:07 ~ 01:41:37 (27.8s) 01:39:14 ~ 01:39:24 (4.9s)

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) 2,836 ms 2,774 ms 3,786 ms 24,542 ms 1,922 ms

首文本(SSE delta) 2,837 ms 2,776 ms 3,788 ms 24,543 ms 2,373 ms

总耗时 4,052 ms 3,308 ms 4,816 ms 24,750 ms 3,676 ms

输入 3 3 3 3 3

输出 59 50 54 42 72

缓存创建

5m: 0 / 1h: 35

5m: 0 / 1h: 41

1,747

5m: 0 / 1h: 1,747

5m: 43 / 1h: 0

缓存读取 32,545 32,562 32,565 30,993 32,562

输出速度 48.6 tok/s 94.0 tok/s 52.5 tok/s - 55.3 tok/s

响应

我是 Claude，由 Anthropic 开发的 AI 助手。我基于大型语言模型构建，旨在提供有益、无害且诚实的对话与任务协助。

我叫 Claude，是 Anthropic 开发的 AI 助手，属于 Claude 产品系列。当前版本为 Claude Sonnet 4.6。

我叫 Claude，是由 Anthropic 开发的 AI 助手，当前使用的模型是 claude-sonnet-4-6，属于 Claude 产品系列。

我是 Claude，由 Anthropic 打造的 AI 助手。当前运行的版本是 claude-sonnet-4-6。

我是 Claude，由 Anthropic 开发的 AI 助手。当前版本为 Claude Sonnet 4.6，擅长代码、写作、分析等多种任务，致力于提供安全、有帮助的回答。

步骤 5: knowledge_recall text

时间 01:42:34 ~ 01:42:45 (5.2s) 01:42:58 ~ 01:43:08 (5.1s) 01:42:05 ~ 01:42:15 (6.7s) 01:43:28 ~ 01:43:44 (14.8s) 01:40:55 ~ 01:41:06 (5.2s)

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) 1,765 ms 1,776 ms 2,576 ms 6,445 ms 1,874 ms

首文本(SSE delta) 1,766 ms 1,777 ms 2,577 ms 6,455 ms 1,875 ms

总耗时 4,200 ms 4,199 ms 5,583 ms 14,023 ms 3,942 ms

输入 3 3 3 3 3

输出 205 205 204 204 205

缓存创建

8,028

5m: 0 / 1h: 8,028

8,042

5m: 0 / 1h: 8,042

8,049

5m: 0 / 1h: 8,049

8,032

5m: 0 / 1h: 8,032

8,066

5m: 8,066 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 84.2 tok/s 84.6 tok/s 67.9 tok/s 27.0 tok/s 99.2 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 01:45:17 ~ 01:45:37 (19.2s) 01:46:30 ~ 01:46:50 (19.0s) 01:44:06 ~ 01:44:31 (21.6s) 01:45:19 ~ 01:45:44 (23.7s) 01:44:26 ~ 01:44:51 (20.7s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=661400470149599196804931671212380177397687591190923428175062677146710725

HTTP 头(TTFB) 1,410 ms 2,141 ms 2,839 ms 7,990 ms 949 ms

首文本(SSE delta) 1,442 ms 2,142 ms 2,840 ms 7,991 ms 1,413 ms

总耗时 18,107 ms 17,932 ms 20,444 ms 22,935 ms 19,294 ms

输入 164 164 164 164 164

输出 1,281 1,623 1,623 1,553 1,536

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 76.9 tok/s 102.8 tok/s 92.2 tok/s 103.9 tok/s 85.9 tok/s

响应

## Counting digits in s

**s** = `661400470149599196804931671212380177397687591190923428175062677146710725`

### Step 1 — Length (groups of 10)

```
6614004701 | 4959919680 | 4931671212 | 3801773976 |

## Counting digits in:
`661400470149599196804931671212380177397687591190923428175062677146710725`

### Length
Grouping by 10s:
`6614004701` `4959919680` `4931671212` `3801773976` `8759119092` `3428175

```python
s = "661400470149599196804931671212380177397687591190923428175062677146710725"
n = len(s)
c0 = s.count('0')
c7 = s.count('7')
c9 = s.count('9')
```

**n = 72**

Digit-by-digit scan:

| pos |

```
s = 661400470149599196804931671212380177397687591190923428175062677146710725
```

**Counting length:** Breaking into groups of 10:
`6614004701` | `4959919680` | `4931671212` | `3801773976` | `8759

## Counting digits in `s`

**s** = `661400470149599196804931671212380177397687591190923428175062677146710725`

### Step 1 — Length

Breaking into groups of 10:

| Chunk | Digits |
|---|---|
| `6614004

总计

总输入(含缓存) 138,157 138,211 138,226 138,755 138,237

总输出 1,580 1,913 1,916 1,834 1,848

总缓存创建

40,608

5m: 0 / 1h: 40,608

40,645

5m: 0 / 1h: 40,645

40,655

5m: 0 / 1h: 40,655

42,480

5m: 0 / 1h: 42,480

73,178

5m: 73,178 / 1h: 0

总缓存读取 97,370 97,387 97,392 96,096 64,880

输出速度 76.1 tok/s 100.2 tok/s 86.9 tok/s 78.0 tok/s 85.3 tok/s

总执行时间 44.9s 40.0s 49.9s 1m 18.2s 46.3s

总墙钟时间

14m 32.0s

含步骤间等待

15m 38.9s

含步骤间等待

13m 28.1s

含步骤间等待

14m 35.0s

含步骤间等待

13m 44.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2057

Anthropic: Claude Sonnet 4.6

$0.2109

Anthropic: Claude Sonnet 4.6

$0.2110

Anthropic: Claude Sonnet 4.6

$0.2162

Anthropic: Claude Sonnet 4.6

$0.2192

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 one-api (100%) direct (100%) one-api (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_01RGyg4ApPYKjr4z7orwaoeV msg_01XayV8myMXXvzTwrmBkuja9 msg_01HUmy41ErmdfT87gHVUc3ux msg_012aXZ2nhoyC4C9cidSiWj14 msg_017K5FRCXaC6QMT4EVXyK75y

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	server: nginx x-new-api-version: 0b8be5cf x-oneapi-request-id: 202606290145184180642740SM0zgDq	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	—
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_011CcW5NDjLoyodLRSbTuYSb
O-Max	cf-ray: a12e7f674965f5da-AMS server: cloudflare x-oneapi-request-id: 2026062901440812044429429e0SzD2	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-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	a12e7f674965f5da-AMS
O-Pro	server: nginx	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	—
O-Bdrk	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606281744276557380018268d9d6CfnGN6TH	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 98 100 97 97 81

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 10.0 /10 ×20 10.0 /10 ×20 5.5 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 0.0 /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 7.7 /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 10.0 /10 ×8 10.0 /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 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 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 10.0 /10 ×5

流式投递 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /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 98 v3.25.0 基准（参考） 97 v3.25.0 97 v3.25.0 81 v3.25.0

雷达图例

共 20 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 档位思考量 tier_thinking_volume_match
20 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 6431 ms 5619 ms 7098 ms 11872 ms 6221 ms

输出速度 76.1 tok/s 100.2 tok/s 86.9 tok/s 78.0 tok/s 85.3 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 ⓘ

$0.2057

$0.2109

$0.2110

$0.2162

$0.2192

缓存读取占比 ⓘ 70% 70% 70% 69% 47%

揭晓

通道

服务商 SudoCode ↗

通道M-Max

官方基准

服务商 ClaudeCN ↗

通道O-Max

服务商 AiTokensFlux ↗

通道O-Pro

服务商 ClaudeAPI ↗

通道O-Bdrk

返回列表