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

2026/06/24 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

官方基准

WorldBase.ai

M-

ByteCatCode

O-Max

ClaudeAPI

O-Bdrk

duckcoding

cc专用

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 08:44:09 ~ 08:44:14 (4.6s) 08:44:07 ~ 08:44:17 (7.2s) 08:44:05 ~ 08:44:15 (5.9s) 08:44:01 ~ 08:44:46 (41.6s) 08:44:03 ~ 08:45:33 (85.6s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,255 ms 5,442 ms 3,007 ms 18,308 ms 3,391 ms

首文本(SSE delta) 3,256 ms 5,469 ms - 39,923 ms 3,392 ms

总耗时 3,368 ms 5,488 ms 4,211 ms 40,032 ms 3,639 ms

输入 3 14 3 3 3

输出 5 2 5 5 5

缓存创建

32,318

5m: 0 / 1h: 32,318

32,611

5m: 0 / 1h: 32,611

32,346

5m: 32,346 / 1h: 0

32,364

5m: 32,364 / 1h: 0

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 08:45:26 ~ 08:45:31 (4.7s) 08:46:22 ~ 08:46:32 (7.2s) 08:46:03 ~ 08:46:13 (5.0s) 08:47:31 ~ 08:47:41 (6.7s) 08:48:45 ~ 08:48:50 (4.7s)

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。自报身份，输出严格遵守以下 3 行格式（不允许 markdown、代码块、额外文本）：
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) 3,242 ms 5,512 ms 3,115 ms 4,940 ms 3,145 ms

首文本(SSE delta) 3,244 ms 5,514 ms - 4,941 ms 3,146 ms

总耗时 3,681 ms 6,012 ms 3,960 ms 5,341 ms 3,576 ms

输入 3 240 3 3 3

输出 22 17 22 22 22

缓存创建

192

5m: 0 / 1h: 192

32,626

5m: 0 / 1h: 32,626

192

5m: 192 / 1h: 0

192

5m: 192 / 1h: 0

192

5m: 0 / 1h: 192

缓存读取 32,318 0 32,346 32,364 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 08:47:47 ~ 08:47:52 (3.3s) 08:50:18 ~ 08:50:33 (10.8s) 08:48:28 ~ 08:48:33 (4.7s) 08:51:28 ~ 08:51:38 (7.1s) 08:51:55 ~ 08:52:00 (3.7s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 2,085 ms 9,384 ms 2,719 ms 5,361 ms 2,354 ms

首文本(SSE delta) 2,195 ms 9,626 ms - 5,518 ms 2,355 ms

总耗时 2,278 ms 9,645 ms 3,496 ms 5,594 ms 2,829 ms

输入 3 17 3 13 3

输出 8 3 8 2 8

缓存创建

5m: 0 / 1h: 38

261

5m: 0 / 1h: 261

5m: 55 / 1h: 0

33,945

5m: 0 / 1h: 57

缓存读取 32,510 32,626 32,538 0 32,510

响应

2025-08

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 08:51:19 ~ 08:51:24 (4.1s) 08:52:45 ~ 08:52:55 (9.5s) 08:52:25 ~ 08:52:35 (5.3s) 08:53:14 ~ 08:53:24 (8.6s) 08:54:57 ~ 08:55:02 (4.2s)

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) 2,131 ms 7,215 ms 2,870 ms 5,775 ms 2,257 ms

首文本(SSE delta) 2,135 ms 7,216 ms - 5,929 ms 2,266 ms

总耗时 3,128 ms 8,287 ms 3,997 ms 7,304 ms 3,054 ms

输入 3 33 3 20 3

输出 62 40 54 63 65

缓存创建

5m: 0 / 1h: 39

1,460

5m: 0 / 1h: 1,460

5m: 43 / 1h: 0

5m: 0 / 1h: 43

缓存读取 32,548 31,446 32,593 33,965 32,567

输出速度 62.4 tok/s 37.3 tok/s 47.9 tok/s 45.8 tok/s 82.5 tok/s

响应

我是 Claude，由 Anthropic 公司打造的 AI 助手。当前版本为 claude-sonnet-4-6，致力于提供安全、有帮助的对话与任务协助。

我是 Claude，由 Anthropic 开发的 AI 助手。我基于 claude-sonnet-4-6 模型运行，致力于提供安全、有益的对话和任务协助。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助回答问题、分析信息、编写代码、创作内容等多种任务。

我是 Claude，一款由 Anthropic 驱动的 AI 开发环境助手。我基于 Claude 模型构建，专注于帮助开发者完成软件工程任务，包括编写代码、调试、重构和系统设计。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助您完成写作、分析、编程、问答等多种任务，致力于提供安全、有帮助的回复。

步骤 5: knowledge_recall text

时间 08:54:11 ~ 08:54:21 (5.5s) 08:55:44 ~ 08:55:55 (5.4s) 08:56:28 ~ 08:56:38 (6.5s) 08:56:24 ~ 08:56:34 (6.9s) 08:56:30 ~ 08:56:40 (5.4s)

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) 1,743 ms 2,170 ms 2,235 ms 3,542 ms 1,842 ms

首文本(SSE delta) 1,751 ms 2,233 ms - 3,572 ms 1,875 ms

总耗时 4,472 ms 4,303 ms 5,211 ms 5,401 ms 4,187 ms

输入 3 7,815 3 659 3

输出 205 159 205 157 204

缓存创建

8,038

5m: 0 / 1h: 8,038

863

5m: 0 / 1h: 863

8,079

5m: 8,079 / 1h: 0

8,047

8,064

5m: 0 / 1h: 8,064

缓存读取 0 0 0 0 0

输出速度 75.3 tok/s 76.8 tok/s 68.9 tok/s 85.8 tok/s 88.2 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 08:56:48 ~ 08:57:08 (16.1s) 08:59:53 ~ 09:00:08 (14.7s) 08:58:38 ~ 08:59:03 (20.7s) 08:59:45 ~ 09:05:50 (360.9s) 08:59:12 ~ 08:59:32 (17.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=675958189597296270582975812071949585435222951051211848649611900938941398

HTTP 头(TTFB) 972 ms 1,671 ms 1,912 ms 13,916 ms 2,755 ms

首文本(SSE delta) 973 ms 1,698 ms - - 3,163 ms

总耗时 15,053 ms 13,380 ms 19,684 ms 13,917 ms 16,865 ms

输入 164 473 193 191 164

输出 1,347 584 1,835 952 1,161

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 95.7 tok/s 50.0 tok/s 103.3 tok/s burst 84.7 tok/s

响应

## Counting digits in s

**s = `675958189597296270582975812071949585435222951051211848649611900938941398`**

### Length

Breaking into groups of 4:
`6759 5818 9597 2962 7058 2975 8120 7194 9585 4352 2

```json
{"n":72,"c":[5,5,13]}
```

## Counting digits in s

**s** = `675958189597296270582975812071949585435222951051211848649611900938941398`

### Length (groups of 10):
```
6759581895 | 9729627058 | 2975812071 | 9495854352 |
22951051

{"n":72,"c":[5,5,13]}

```json
{"n":72,"c":[5,5,13]}
```

**Verification:**

String: `675958189597296270582975812071949585435222951051211848649611900938941398`

**Length:** Split into groups of 10 → 7×10 + 2 = **72**

**0s*

总计

总输入(含缓存) 138,180 140,485 138,400 141,782 138,248

总输出 1,649 805 2,129 1,201 1,465

总缓存创建

40,625

5m: 0 / 1h: 40,625

67,821

5m: 0 / 1h: 67,821

40,715

5m: 40,715 / 1h: 0

74,564

5m: 32,556 / 1h: 0

40,674

5m: 0 / 1h: 40,674

总缓存读取 97,376 64,072 97,477 66,329 97,395

输出速度 90.7 tok/s 52.8 tok/s 95.7 tok/s 68.7 tok/s 85.1 tok/s

总执行时间 38.3s 54.8s 48.1s 7m 11.8s 2m 1.6s

总墙钟时间

12m 59.2s

含步骤间等待

16m 1.0s

含步骤间等待

14m 58.1s

含步骤间等待

21m 49.7s

含步骤间等待

15m 29.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2068

Anthropic: Claude Sonnet 4.6

$0.3114

Anthropic: Claude Sonnet 4.6

$0.1572

Anthropic: Claude Sonnet 4.6

$0.1169

Anthropic: Claude Sonnet 4.6

$0.2043

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 - - -

ID 格式 msg_01Mvi2waew7JStdjpymrk7GK msg_545208a121f7491bbeeaca8ff60e6ef5 msg_01CzSc1Ftbig7pcN8D1jnPVH msg_01XwZFmyUPB1FRFkqLad9RL8 msg_01QQVmVD3MdStwjJDSYvmWNB

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
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_011CcMB8og3CACgaAtUvSr5y
M-	cf-ray: a107cae23f23fc28-AMS server: cloudflare	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:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	b6631490-e614-459a-862c-7c7d506d861d a107cae23f23fc28-AMS
O-Max	server: nginx/1.29.8 x-oneapi-request-id: 202606240058393592892248268d9d69FF3pgyD	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-Bdrk	via: 1.1 Caddy server: nginx set-cookie: server_name_session=3d5ebdb2816e1f946d236297cd497bd0; Max-Age=86400; httponly; path=/ x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606240105323429662878268d9d6WMlS19Rk	id_format: msg_prefix via_chain: ["1.1 Caddy"] extra_fields: ["context_management","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:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	07fd4139-9db4-4006-a658-38c283dbdfa3
cc专用	server: Nginx x-oneapi-request-id: 202606240059149909951988268d9d6MwxcqKjM	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	—

协议指纹评分（机器自动）

总分 / 100 100 84 89 75 97

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 5.0 /10 ×20 10.0 /10 ×20 5.5 /10 ×20 10.0 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 10.0 /10 ×15 0.0 /10 ×15 0.0 /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 7.7 /10 ×13 10.0 /10 ×13 7.7 /10 ×13 10.0 /10 ×13

原生 msg-ID 10.0 /10 ×8 0.0 /10 ×8 10.0 /10 ×8 4.0 /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 4.0 /10 ×6 10.0 /10 ×6

inference_geo 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 4.0 /10 ×5 10.0 /10 ×5

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

流式投递 10.0 /10 ×5 10.0 /10 ×5 未计入 10.0 /10 ×5 10.0 /10 ×5

Req-ID 透传 10.0 /10 ×4 0.0 /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

与基线相似度基准（参考） 84 v3.24.0 89 v3.24.0 75 v3.24.0 97 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 5330 ms 7853 ms 6760 ms 12932 ms 5692 ms

输出速度 90.7 tok/s 52.8 tok/s 95.7 tok/s 68.7 tok/s 85.1 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 ⓘ

$0.2068

$0.3114

$0.1572

$0.1169

$0.2043

缓存读取占比 ⓘ 70% 46% 70% 47% 70%

揭晓

通道

官方基准

服务商 WorldBase.ai ↗

通道M-

服务商 ByteCatCode ↗

通道O-Max

服务商 ClaudeAPI ↗

通道O-Bdrk

服务商 duckcoding ↗

通道cc专用

返回列表