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

2026/06/21 运行日期

claude-sonnet-4-6 官方基准

3 通道分数 < 70 另含 1 任务失败

已揭晓揭晓进度

步骤 / 指标

SSSAiCode

O-Max-L2

官方基准

FoxCode

R-ultra

SSSAiCode

R-Kiro

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 23:53:26 ~ 23:53:31 (3.2s) 23:53:34 ~ 23:53:39 (4.8s) 23:53:32 ~ 23:53:37 (4.0s) 23:53:28 ~ 23:53:33 (3.4s) 23:53:30 ~ 23:53:45 (10.4s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,123 ms 3,326 ms 2,230 ms 2,348 ms 8,785 ms

首文本(SSE delta) 2,230 ms 3,327 ms 2,231 ms 2,458 ms 8,786 ms

总耗时 2,298 ms 3,505 ms 2,985 ms 2,473 ms 8,787 ms

输入 3 3 60 0 3

输出 5 5 15 2 5

缓存创建

31,991

5m: 0 / 1h: 31,991

32,318

5m: 0 / 1h: 32,318

32,956

5m: 32,956 / 1h: 0

26,613

5m: 0 / 1h: 26,613

缓存读取 0 0 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 23:55:45 ~ 23:55:50 (3.3s) 23:57:27 ~ 23:57:32 (3.7s) 23:54:53 ~ 23:54:58 (4.3s) 23:55:26 ~ 23:55:31 (3.4s) 23:55:12 ~ 23:55:22 (8.9s)

提示词

查看本步 Prompt

以下是对你自身的 metadata 探针。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。按以下 3 行回答自己的身份，每行一个 label，不要寒暄不要解释：
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) 2,235 ms 2,089 ms 3,032 ms 2,316 ms 7,671 ms

首文本(SSE delta) 2,506 ms 2,092 ms 3,033 ms 2,322 ms 7,673 ms

总耗时 2,565 ms 2,714 ms 3,317 ms 2,545 ms 7,673 ms

输入 3 3 21 0 3

输出 22 22 82 17 22

缓存创建

202

5m: 0 / 1h: 202

202

5m: 0 / 1h: 202

33,526

5m: 33,526 / 1h: 0

202

5m: 0 / 1h: 202

缓存读取 31,991 32,318 0 0 26,613

响应

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

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

步骤 3: cutoff text

时间 23:59:41 ~ 23:59:46 (2.6s) 06/22 00:00:55 ~ 00:01:05 (5.4s) 23:58:12 ~ 23:58:17 (3.8s) 23:57:56 ~ 23:58:01 (3.2s) 23:58:10 ~ 23:58:15 (4.2s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 1,360 ms 3,689 ms 2,494 ms 2,194 ms 2,956 ms

首文本(SSE delta) 1,513 ms 3,809 ms 2,669 ms 2,399 ms 2,959 ms

总耗时 1,643 ms 4,073 ms 2,925 ms 2,402 ms 2,959 ms

输入 3 3 67 0 2

输出 8 8 36 3 6

缓存创建

5m: 0 / 1h: 38

1,723

5m: 0 / 1h: 1,723

33,884

5m: 33,884 / 1h: 0

38,985

5m: 0 / 1h: 38,985

缓存读取 32,193 30,854 0 0 0

输出速度 - - 140.6 tok/s - -

响应

2025-04

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 06/22 00:00:48 ~ 00:01:08 (15.5s) 06/22 00:02:26 ~ 00:02:31 (3.9s) 06/22 00:00:50 ~ 00:01:00 (5.3s) 06/22 00:01:10 ~ 00:01:15 (4.8s) 06/22 00:00:30 ~ 00:00:40 (7.5s)

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) 11,126 ms 1,872 ms 2,453 ms 2,646 ms 5,802 ms

首文本(SSE delta) 11,636 ms 1,874 ms 2,454 ms 2,728 ms 5,803 ms

总耗时 11,729 ms 2,839 ms 4,145 ms 3,980 ms 5,803 ms

输入 3 3 82 0 3

输出 47 58 54 38 67

缓存创建

5m: 0 / 1h: 35

350

5m: 350 / 1h: 0

1,764

5m: 0 / 1h: 1,764

缓存读取 32,231 32,577 33,884 0 25,149

输出速度 - 60.1 tok/s 31.9 tok/s 30.4 tok/s burst

响应

我是 Claude，由 Anthropic 开发的 AI 助手。Anthropic 是一家专注于 AI 安全研究的美国公司。

我是 Claude，由 Anthropic 开发的 AI 助手。当前版本为 claude-sonnet-4-6，致力于提供安全、有益的对话和任务协助。

我是 Claude，由 Anthropic 打造的 AI 助手。当前运行的模型版本为 claude-sonnet-4-6。

我是 Kiro，一个由 Anthropic 开发的 AI 驱动开发环境。我基于 Claude 模型构建，专注于帮助开发者完成软件工程任务，让开发者专注于系统设计和决策。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助回答问题、分析信息、编写代码、创作内容等，致力于做一个安全、有益的智能助手。

步骤 5: knowledge_recall text

时间 06/22 00:02:55 ~ 00:03:00 (4.7s) 06/22 00:04:41 ~ 00:04:46 (4.6s) 06/22 00:02:49 ~ 00:03:20 (28.2s) 06/22 00:03:10 ~ 00:03:15 (4.4s) 06/22 00:03:26 ~ 00:03:46 (15.4s)

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) 1,618 ms 1,595 ms 1,289 ms 1,525 ms 14,192 ms

首文本(SSE delta) 3,714 ms 1,597 ms 1,290 ms 1,568 ms 14,193 ms

总耗时 3,772 ms 3,596 ms 27,172 ms 3,518 ms 14,193 ms

输入 3 3 82 0 3

输出 198 204 2,283 223 804

缓存创建

7,702

5m: 0 / 1h: 7,702

8,059

5m: 0 / 1h: 8,059

9,591

5m: 9,591 / 1h: 0

19,293

5m: 0 / 1h: 19,293

缓存读取 0 0 0 0 0

输出速度 - 102.1 tok/s 88.2 tok/s 114.4 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 06/22 00:06:49 ~ 00:06:59 (6.7s) 06/22 00:08:14 ~ 00:08:34 (19.6s) 06/22 00:05:11 ~ 00:05:41 (28.2s) 06/22 00:06:10 ~ 00:06:35 (21.6s) 06/22 00:12:00 ~ 00:19:15 (434.4s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=303854346484597211763593923146592817223869786935933424773917398994378196

HTTP 头(TTFB) - 1,224 ms 7,001 ms 2,370 ms 30,956 ms

首文本(SSE delta) - 1,226 ms 7,003 ms 2,405 ms 30,957 ms

总耗时 - 18,518 ms 27,125 ms 20,672 ms 30,957 ms

输入 - 164 90 0 3

输出 - 1,669 2,401 1,503 2,019

缓存创建

5m: 0 / 1h: 0

119

5m: 119 / 1h: 0

17,072

5m: 0 / 1h: 17,072

缓存读取 - 0 0 0 0

输出速度 - 96.5 tok/s 119.3 tok/s 82.3 tok/s burst

响应

失败

评测失败

## Counting digits in s

**s = `303854346484597211763593923146592817223869786935933424773917398994378196`**

### Length Count (chunking by 10):
```
3038543464  → 10
8459721176  → 10
3593923146  → 10
5

{"n":72,"c":[1,8,12]}

```
s = 303854346484597211763593923146592817223869786935933424773917398994378196
```

Indexing each character (1-based):

Position map:
```
 1: 3   2: 0   3: 3   4: 8   5: 5   6: 4   7: 3   8: 4   9:

## Counting digits in `s`

**s** = `303854346484597211763593923146592817223869786935933424773917398994378196`

I'll index every character:

| Pos | Digit | Pos | Digit | Pos | Digit | Pos | Digit |
|-

总计

总输入(含缓存)

136,398

仅前 6 步

138,265

144,712

155,708

总输出

280

仅前 6 步

1,966

4,871

1,786

2,923

总缓存创建

39,968

5m: 0 / 1h: 39,968

仅前 6 步

42,337

5m: 0 / 1h: 42,337

110,426

5m: 110,426 / 1h: 0

103,929

5m: 0 / 1h: 103,929

总缓存读取

96,415

仅前 6 步

95,749

33,884

51,762

输出速度

仅前 6 步

95.3 tok/s

99.6 tok/s

82.2 tok/s

burst

总执行时间

36.0s

仅前 6 步

41.9s

1m 13.8s

40.8s

8m 0.8s

总墙钟时间

13m 33.0s

含步骤间等待

仅前 6 步

15m 0.0s

含步骤间等待

12m 9.1s

含步骤间等待

13m 7.0s

含步骤间等待

25m 45.9s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1830

Anthropic: Claude Sonnet 4.6 (部分)

$0.2175

Anthropic: Claude Sonnet 4.6

$0.3432

Anthropic: Claude Sonnet 4.6

>=$0.0268

Anthropic: Claude Sonnet 4.6 (成本不完整)

$0.4492

Anthropic: Claude Sonnet 4.6

通道指纹

来源步骤 knowledge_recall digit_count digit_count digit_count digit_count

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

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

CDN - - cloudflare - -

ID 格式 msg_016gxa6u52zVcC3SBXdY2L8C msg_01FWGNWUfgDPAJ4bm5EQ2ZVw msg_01bSBmyrWieGBcl9Jo9YLFry msg_6872c620405241199e82a59ab11e94c1 msg_01KeEvc8HZ7bvjNYDPAeLSX1

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-L2	cf-ray: a0f43d93bfcf8ad4-LAX server: Photon-Edge set-cookie: _cfuvid=1pb3wYLeBeTZhtmuGFceDE3bjy4PBiqRno750SkME0I-1782057777.2408533-1.0.1.1-gpv1n7yhazBNofn4b7JNwaJZaaJ4bmBjEuyN.0q5MR0; 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	1782057777193-g6q08k0wc5sin req_011CcGgojo56QT4Pzco8or5W a0f43d93bfcf8ad4-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_011CcGhDBDB5omd4PVAJK6BV
R-ultra	cf-ray: a0f440dfcde3e393-NRT server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606211605124555033468268d9d6TxUmZnzq	id_format: msg_prefix 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_fieldsupstream:anthropic-direct-from-body	a0f440dfcde3e393-NRT
R-Kiro	server: Photon-Edge x-amzn-requestid: a5291e5c-d702-4b09-9bc2-ed6b84f09923	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	1782057971129-rtyn1nif0vbfs a5291e5c-d702-4b09-9bc2-ed6b84f09923
u-cc	server: nginx x-new-api-version: 1.0.14 x-oneapi-request-id: 202606211618437600778618268d9d65hiiiK5h	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 / 100 100 73 52 71

基线状态已对照基线已对照基线已对照基线已对照基线

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 5.0 /10 ×14 6.7 /10 ×14

sliding 5m — 10.0 /10 ×13 5.3 /10 ×13 5.0 /10 ×13 7.7 /10 ×13

缓存 TTL 一致 — 10.0 /10 ×15 0.0 /10 ×15 0.0 /10 ×15 10.0 /10 ×15

缓存命中比 — 10.0 /10 ×20 0.0 /10 ×20 0.0 /10 ×20 1.0 /10 ×20

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

知识截止 — 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 0.0 /10 ×7 10.0 /10 ×7

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

Req-ID 透传 — 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

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

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

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

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

延迟基线 — 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 0.0 /10 ×5

综合结论

与基线相似度执行失败，不评分基准（参考） 73 v3.24.0 52 v3.24.0 71 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 4401 ms 5874 ms 11278 ms 5932 ms 11729 ms

输出速度 - 95.3 tok/s 99.6 tok/s 82.2 tok/s burst

完成步骤 5/6 成功 · 1 失败 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 ⓘ

$0.1830 (部分)

$0.2175

$0.3432

>=$0.0268 (成本不完整)

$0.4492

缓存读取占比 ⓘ 71% (部分) 69% 23% - 33%

揭晓

通道

服务商 SSSAiCode ↗

通道O-Max-L2

官方基准

服务商 FoxCode ↗

通道R-ultra

服务商 SSSAiCode ↗

通道R-Kiro

服务商 HongMaCC ↗

通道u-cc

返回列表