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

2026/06/30 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

duckcoding

cc专用

LinkAPI

O-Max

官方基准

SSSAiCode

R-Kiro

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 03:56:26 ~ 03:56:36 (8.8s) 03:56:22 ~ 03:56:27 (4.3s) 03:56:28 ~ 03:56:33 (3.8s) 03:56:24 ~ 03:56:39 (12.8s) 03:56:20 ~ 03:56:30 (8.2s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 6,573 ms 2,933 ms 2,652 ms 8,760 ms 6,522 ms

首文本(SSE delta) 6,703 ms 2,935 ms 2,654 ms 8,856 ms 6,528 ms

总耗时 7,415 ms 3,034 ms 2,774 ms 8,956 ms 6,684 ms

输入 3 3 3 0 3

输出 5 5 5 2 5

缓存创建

32,318

5m: 0 / 1h: 32,318

31,364

5m: 31,364 / 1h: 0

32,318

5m: 0 / 1h: 32,318

32,318

5m: 0 / 1h: 32,318

缓存读取 0 0 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 03:58:47 ~ 03:58:52 (4.1s) 03:58:57 ~ 03:59:02 (4.4s) 03:58:09 ~ 03:58:14 (3.1s) 03:58:14 ~ 03:58:19 (3.8s) 03:59:40 ~ 03:59:50 (5.5s)

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) 2,594 ms 3,040 ms 1,798 ms 2,527 ms 3,498 ms

首文本(SSE delta) 2,887 ms 3,041 ms 1,799 ms 2,548 ms 3,502 ms

总耗时 3,056 ms 3,311 ms 2,165 ms 2,923 ms 3,869 ms

输入 3 3 3 0 3

输出 22 22 22 18 22

缓存创建

202

5m: 0 / 1h: 202

202

5m: 202 / 1h: 0

202

5m: 0 / 1h: 202

31,566

5m: 31,566 / 1h: 0

缓存读取 32,318 31,364 32,318 0 0

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 04:00:31 ~ 04:00:36 (3.8s) 04:01:56 ~ 04:02:06 (5.1s) 04:01:00 ~ 04:01:05 (3.6s) 04:01:02 ~ 04:01:07 (4.2s) 04:01:35 ~ 04:01:40 (4.7s)

提示词

查看本步 Prompt

请直接回答你的知识截止日期，格式 YYYY-MM，不要附加任何说明。

HTTP 头(TTFB) 2,667 ms 3,924 ms 2,324 ms 2,772 ms 3,050 ms

首文本(SSE delta) 2,671 ms 3,925 ms 2,508 ms 2,991 ms 3,051 ms

总耗时 2,925 ms 3,928 ms 2,634 ms 3,144 ms 3,310 ms

输入 3 3 3 0 3

输出 8 8 8 3 8

缓存创建

5m: 0 / 1h: 57

5m: 38 / 1h: 0

5m: 0 / 1h: 55

5m: 57 / 1h: 0

缓存读取 32,520 31,566 32,520 0 31,566

响应

2025-08

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 04:03:44 ~ 04:03:49 (4.9s) 04:05:35 ~ 04:05:45 (5.9s) 04:02:15 ~ 04:02:20 (3.5s) 04:02:28 ~ 04:02:33 (4.8s) 04:03:29 ~ 04:03:39 (5.5s)

提示词

查看本步 Prompt

中文回答：你叫什么、属于哪个产品、谁开发？50 字内。

HTTP 头(TTFB) 2,977 ms 3,484 ms 1,885 ms 2,567 ms 3,505 ms

首文本(SSE delta) 3,322 ms 3,485 ms 1,887 ms 2,864 ms 3,506 ms

总耗时 3,531 ms 4,763 ms 2,533 ms 3,991 ms 4,080 ms

输入 3 3 3 0 3

输出 41 69 48 40 46

缓存创建

5m: 0 / 1h: 41

5m: 39 / 1h: 0

5m: 0 / 1h: 43

298

5m: 0 / 1h: 298

缓存读取 32,577 31,604 32,575 0 32,318

输出速度 - 54.0 tok/s 74.3 tok/s 35.5 tok/s 80.1 tok/s

响应

我叫 Claude，是由 Anthropic 开发的 AI 助手，当前版本为 claude-sonnet-4-6。

我是 Claude，由 Anthropic 公司打造的 AI 助手。目前运行的模型版本为 claude-sonnet-4-6，作为 Claude Code CLI 工具为你提供软件工程相关帮助。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助完成写作、分析、编程、问答等各类任务。

我是 Kiro，一款由 Anthropic 开发的 AI 驱动开发环境。我基于 Claude 模型构建，专注于协助开发者完成软件工程任务，让开发者专注于设计和决策。

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

步骤 5: knowledge_recall text

时间 04:07:46 ~ 04:07:56 (5.9s) 04:09:04 ~ 04:09:24 (16.0s) 04:05:33 ~ 04:05:38 (4.8s) 04:03:37 ~ 04:03:42 (4.8s) 04:07:20 ~ 04:07:35 (14.8s)

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) 1,828 ms 1,958 ms 1,986 ms 1,953 ms 3,427 ms

首文本(SSE delta) 1,830 ms 3,228 ms 1,988 ms 2,099 ms 4,696 ms

总耗时 4,760 ms 14,994 ms 3,896 ms 4,030 ms 13,455 ms

输入 3 3 3 0 3

输出 205 938 204 222 758

缓存创建

8,048

5m: 0 / 1h: 8,048

8,761

5m: 8,761 / 1h: 0

8,055

5m: 0 / 1h: 8,055

8,757

5m: 8,757 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 70.0 tok/s 79.7 tok/s 106.9 tok/s 115.0 tok/s 86.5 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 04:09:56 ~ 04:10:16 (19.0s) 04:12:26 ~ 04:12:41 (12.8s) 04:06:53 ~ 04:07:13 (18.9s) 04:05:46 ~ 04:06:11 (21.8s) 04:08:59 ~ 04:09:19 (15.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=805431601888310645116077323689956521176473546499820816025930084984891522

HTTP 头(TTFB) 2,214 ms 1,254 ms 755 ms 2,361 ms 3,203 ms

首文本(SSE delta) 2,535 ms 2,417 ms 757 ms 2,413 ms 3,257 ms

总耗时 18,249 ms 11,869 ms 17,962 ms 21,115 ms 14,673 ms

输入 164 3 164 0 164

输出 1,502 1,014 1,813 1,311 931

缓存创建

5m: 0 / 1h: 0

1,924

5m: 1,924 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 95.6 tok/s 107.3 tok/s 105.4 tok/s 70.1 tok/s 81.6 tok/s

响应

## Counting s = `805431601888310645116077323689956521176473546499820816025930084984891522`

**Indexing each character:**

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

```json
{"n":72,"c":[8,4,7]}
```

**Parsing** `s = 805431601888310645116077323689956521176473546499820816025930084984891522`

**Length:**
```
8 0 5 4 3 1 6 0 1 8  → 10
8 8 3 1 0 6 4 5 1 1  → 20
6 0 7 7 3 2 3 6 8 9  → 30
9 5 6 5 2 1 1

Counting the string character by character:

s = `805431601888310645116077323689956521176473546499820816025930084984891522`

Indexing (1-based):
```
 1: 8   2: 0   3: 5   4: 4   5: 3   6: 1   7: 6   8

```json
{"n":72,"c":[8,4,7]}
```

**Verification walkthrough:**

```
s = 805431601888310645116077323689956521176473546499820816025930084984891522
```

**Length:** Grouped in 10s → 70 + 2 = **72**

**0

总计

总输入(含缓存) 138,260 136,880 138,265 0 137,059

总输出 1,783 2,056 2,100 1,596 1,770

总缓存创建

40,666

5m: 0 / 1h: 40,666

42,328

5m: 42,328 / 1h: 0

40,673

5m: 0 / 1h: 40,673

72,996

5m: 40,380 / 1h: 32,616

总缓存读取 97,415 94,534 97,413 0 63,884

输出速度 91.6 tok/s 89.8 tok/s 104.5 tok/s 72.3 tok/s 83.6 tok/s

总执行时间 46.5s 48.4s 37.8s 52.2s 54.5s

总墙钟时间

13m 49.9s

含步骤间等待

16m 19.0s

含步骤间等待

10m 44.9s

含步骤间等待

9m 47.0s

含步骤间等待

12m 59.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2090

Anthropic: Claude Sonnet 4.6

$0.1585

Anthropic: Claude Sonnet 4.6

$0.2138

Anthropic: Claude Sonnet 4.6

>=$0.0239

Anthropic: Claude Sonnet 4.6 (成本不完整)

$0.2632

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

ID 格式 msg_01QTNCEUT9sWGkZgwD87pdGs msg_0195cs2XhQDSH5eFMhjSYEFQ msg_018WbKCpbsaAqy7tkdFkADyY msg_6aadd9fe5ac342c6b3076b93335230e4 msg_01QxAaaCeJo9RTgF5Q8XLmN8

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
cc专用	server: Nginx x-oneapi-request-id: 202606292009574041319968268d9d6F1YvpMsx	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-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
O-Max	server: nginx set-cookie: server_name_session=7d88fca8b1426ca76c58d5aeba6fb224; Max-Age=86400; httponly; path=/ x-new-api-version: v1.0.0-rc.15 x-oneapi-request-id: 202606292012273298743728268d9d6PwIelM9T	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_011CcY9t2K4SLtqq8KUkSB6x
R-Kiro	server: Photon-Edge x-amzn-requestid: 6cb3fee2-8105-4ce3-b9e6-3a87e933659a	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	1782763547219-07s9hvb3imo8d 6cb3fee2-8105-4ce3-b9e6-3a87e933659a
O-Bdrk	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606292009007004974468268d9d67UUSaav0	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 86 99 50 85

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 0.0 /10 ×20 5.4 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 0.0 /10 ×15 10.0 /10 ×15 0.0 /10 ×15 4.0 /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 5.0 /10 ×14 10.0 /10 ×14

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

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

系统提示纯净 10.0 /10 ×8 5.0 /10 ×8 10.0 /10 ×8 未计入 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 0.0 /10 ×7 10.0 /10 ×7

service_tier 10.0 /10 ×6 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 10.0 /10 ×5 0.0 /10 ×5 10.0 /10 ×5

延迟基线 9.9 /10 ×5 9.3 /10 ×5 10.0 /10 ×5 9.9 /10 ×5 8.0 /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 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

综合结论

与基线相似度 98 v3.25.0 86 v3.25.0 基准（参考） 50 v3.25.0 85 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 6656 ms 6983 ms 5327 ms 7360 ms 7679 ms

输出速度 91.6 tok/s 89.8 tok/s 104.5 tok/s 72.3 tok/s 83.6 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 ⓘ

$0.2090

$0.1585

$0.2138

>=$0.0239 (成本不完整)

$0.2632

缓存读取占比 ⓘ 70% 69% 70% - 47%

揭晓

通道

服务商 duckcoding ↗

通道cc专用

服务商 LinkAPI ↗

通道O-Max

官方基准

服务商 SSSAiCode ↗

通道R-Kiro

服务商 ClaudeAPI ↗

通道O-Bdrk

返回列表