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

2026/06/17 运行日期

claude-sonnet-4-6 官方基准

1 通道分数 < 70 另含 1 任务失败

已揭晓揭晓进度

步骤 / 指标

PackyCode

cc-Zeus

SSSAiCode

O-Max

luminai

M-Kiro/Antigravity

官方基准

RunAPI

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 12:36:39 ~ 12:36:49 (6.2s) 12:36:35 ~ 12:36:40 (3.1s) 12:36:37 ~ 12:36:47 (6.8s) 12:36:48 ~ 12:36:58 (4.8s) 12:36:41 ~ 12:36:51 (7.9s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,451 ms 1,799 ms 5,193 ms 3,059 ms 4,198 ms

首文本(SSE delta) 2,452 ms 1,887 ms 5,195 ms 3,060 ms 4,499 ms

总耗时 2,592 ms 2,036 ms 5,246 ms 3,537 ms 4,677 ms

输入 3 3 1,443 3 3

输出 5 5 1 5 5

缓存创建

32,318

5m: 0 / 1h: 32,318

31,987

5m: 0 / 1h: 31,987

4,111

5m: 4,111 / 1h: 0

5m: 0 / 1h: 0

32,320

5m: 0 / 1h: 32,320

缓存读取 0 0 23,297 32,318 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 12:39:26 ~ 12:39:31 (4.3s) 12:40:20 ~ 12:40:25 (2.5s) 12:39:24 ~ 12:39:34 (6.9s) 12:40:14 ~ 12:40:24 (5.5s) 12:38:34 ~ 12:38:39 (4.1s)

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) 2,029 ms 1,432 ms 5,149 ms 3,862 ms 1,968 ms

首文本(SSE delta) 2,030 ms 1,700 ms 5,150 ms 3,864 ms 1,969 ms

总耗时 2,393 ms 1,770 ms 5,209 ms 4,382 ms 3,005 ms

输入 3 3 1,452 3 3

输出 22 22 24 22 22

缓存创建

192

5m: 0 / 1h: 192

192

5m: 0 / 1h: 192

4,138

5m: 4,138 / 1h: 0

192

5m: 0 / 1h: 192

192

5m: 0 / 1h: 192

缓存读取 32,318 31,987 23,449 32,318 32,320

响应

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

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 12:41:33 ~ 12:41:38 (3.4s) 12:44:01 ~ 12:44:07 (3.3s) 12:41:15 ~ 12:41:25 (7.9s) 12:42:30 ~ 12:42:35 (4.2s) 12:40:12 ~ 12:40:17 (4.8s)

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) 2,213 ms 2,259 ms 6,626 ms 2,772 ms 2,175 ms

首文本(SSE delta) 2,214 ms 2,304 ms 6,797 ms 2,774 ms 2,176 ms

总耗时 2,460 ms 2,390 ms 6,851 ms 3,250 ms 2,499 ms

输入 3 3 1,456 3 3

输出 8 8 4 8 8

缓存创建

5m: 0 / 1h: 57

4,147

5m: 4,147 / 1h: 0

5m: 0 / 1h: 55

5m: 0 / 1h: 57

缓存读取 32,510 32,179 23,500 32,510 32,512

响应

2025-08

2025-04

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 12:45:18 ~ 12:45:23 (4.5s) 12:46:40 ~ 12:46:45 (3.3s) 12:42:48 ~ 12:42:58 (8.1s) 12:46:33 ~ 12:46:38 (4.2s) 12:42:10 ~ 12:42:15 (3.8s)

提示词

查看本步 Prompt

请用中文介绍你自己是什么产品或助手，由谁开发，50 字以内。

HTTP 头(TTFB) 2,053 ms 1,433 ms 5,971 ms 2,570 ms 1,855 ms

首文本(SSE delta) 2,054 ms 2,377 ms 6,937 ms 2,572 ms 1,856 ms

总耗时 3,005 ms 2,610 ms 7,006 ms 3,177 ms 2,690 ms

输入 3 3 1,458 3 3

输出 57 67 65 41 19

缓存创建

5m: 0 / 1h: 43

4,153

5m: 4,153 / 1h: 0

5m: 0 / 1h: 39

5m: 0 / 1h: 35

缓存读取 32,567 32,236 23,539 32,565 32,569

输出速度 59.9 tok/s - - 67.8 tok/s -

响应

我是 Claude，由 Anthropic 开发的 AI 助手，擅长对话、写作、分析、编程等任务，致力于提供安全、有用的帮助。

我是 Claude，由 Anthropic 开发的 AI 助手。我被设计用于对话、写作、分析、编程等多种任务，致力于提供有帮助、无害且诚实的回应。

我是 Claude Sonnet 4.6，由 Anthropic 开发的 AI 助手。我擅长代码编写、分析推理和文本创作，旨在帮助用户高效完成各类任务。

我是 Claude，由 Anthropic 公司打造的 AI 助手。当前版本为 claude-sonnet-4-6。

我是Claude，由Anthropic开发的AI助手。

步骤 5: knowledge_recall text

时间 12:46:53 ~ 12:47:03 (8.7s) 12:50:08 ~ 12:50:18 (5.1s) 12:45:52 ~ 12:46:02 (6.5s) 12:48:21 ~ 12:48:31 (7.3s) 12:44:40 ~ 12:44:50 (5.7s)

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) 1,960 ms 1,909 ms 3,366 ms 2,293 ms 2,261 ms

首文本(SSE delta) 1,961 ms 4,221 ms 5,234 ms 2,294 ms 2,264 ms

总耗时 4,025 ms 4,287 ms 5,292 ms 6,234 ms 4,721 ms

输入 3 3 376 3 3

输出 205 198 211 205 205

缓存创建

8,056

5m: 0 / 1h: 8,056

7,735

5m: 0 / 1h: 7,735

1,071

5m: 1,071 / 1h: 0

2,316

5m: 0 / 1h: 2,316

8,012

5m: 0 / 1h: 8,012

缓存读取 0 0 6,069 5,718 0

输出速度 99.3 tok/s - burst 52.0 tok/s 83.4 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 12:48:09 ~ 12:48:25 (12.1s) 12:52:11 ~ 12:52:16 (1.0s) 12:47:10 ~ 12:47:20 (5.7s) 12:52:28 ~ 12:52:43 (11.0s) 12:47:24 ~ 12:47:49 (20.7s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=048573037342298842425376227356601093173046239648034855856211899345746340

HTTP 头(TTFB) 1,529 ms - 3,683 ms 897 ms 2,027 ms

首文本(SSE delta) 1,530 ms - 3,982 ms 898 ms 3,361 ms

总耗时 11,063 ms - 4,033 ms 10,080 ms 18,870 ms

输入 164 - 190 164 164

输出 701 - 14 822 1,356

缓存创建

5m: 0 / 1h: 0

缓存读取 0 - - 0 0

输出速度 73.5 tok/s - - 89.5 tok/s 87.4 tok/s

响应

```json
{"n":72,"c":[7,6,5]}
```

失败

评测失败

{"n":70,"c":[6,5,4]}

```json
{"n":72,"c":[7,6,5]}
```

**Verification:**

String: `048573037342298842425376227356601093173046239648034855856211899345746340`

- **Length (n):** 72 characters

- **0s** at indices 0,6,31,33,

## Counting digits in s

**s** = `048573037342298842425376227356601093173046239648034855856211899345746340`

### Length
Counting in groups of 10:
`0485730373` | `4229884242` | `5376227356` | `60109317

总计

总输入(含缓存)

138,240

136,431

仅前 6 步

123,849

138,210

138,196

总输出

998

300

仅前 6 步

319

1,103

1,615

总缓存创建

40,666

5m: 0 / 1h: 40,666

40,014

5m: 0 / 1h: 40,014

仅前 6 步

17,620

5m: 17,620 / 1h: 0

2,602

5m: 0 / 1h: 2,602

40,616

5m: 0 / 1h: 40,616

总缓存读取

97,395

96,402

仅前 6 步

99,854

135,429

97,401

输出速度

76.7 tok/s

仅前 6 步

burst

77.8 tok/s

86.9 tok/s

总执行时间

39.1s

18.3s

仅前 6 步

41.9s

37.1s

46.9s

总墙钟时间

11m 45.7s

含步骤间等待

15m 41.1s

含步骤间等待

仅前 6 步

10m 43.5s

含步骤间等待

15m 55.4s

含步骤间等待

11m 8.3s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1972

Anthropic: Claude Sonnet 4.6

$0.1835

Anthropic: Claude Sonnet 4.6 (部分)

$0.0952

Anthropic: Claude Sonnet 4.6

$0.0675

Anthropic: Claude Sonnet 4.6

$0.2063

Anthropic: Claude Sonnet 4.6

通道指纹

来源步骤 digit_count knowledge_recall 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 - cloudflare - -

ID 格式 msg_01NSU3oJ8yKTsFUd1B8UQogQ msg_013X4iC79pE4bbvTYqjkryqz msg_01i7VcnkNbEeudj4wjMfgx7wN9 msg_01Wp9aMoooiEAVed3QT6F1AA msg_01UaqeyW1tZpNaw1uAEVtiSy

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-Zeus	via: 1.1 Caddy cf-ray: a0cf6ba56a4dcd1f-LAX server: cloudflare x-oneapi-request-id: 20260617124811125717713JSJqZm4X	id_format: msg_prefix via_chain: ["1.1 Caddy"] 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-rayhdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a0cf6ba56a4dcd1f-LAX
O-Max	cf-ray: a0cf6e890f42e9dd-LAX server: Photon-Edge set-cookie: _cfuvid=JjiI2ufOcprPWS_DMQvFIk961XNmKybfpEehpiq9bhQ-1781671809.4496117-1.0.1.1-60X0zZGvKWujtvxMOS765v8slmS72T4oUSYOaV3.XZs; 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	1781671809400-wckxcx4pvpf4k req_011Cc8EGPnNk2MDKnY2VnuXA a0cf6e890f42e9dd-LAX
M-Kiro/Antigravity	cf-ray: a0cf6a379a1b312a-AMS server: cloudflare x-oneapi-request-id: 20260617044713398262318268d9d6zlXllTFD	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	a0cf6a379a1b312a-AMS
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_011Cc8ESkBzGe1WCU3gg43qy
O-Max	server: nginx/1.24.0 (Ubuntu) x-oneapi-request-id: 20260617124726473187649DOJKpX4C	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 93 79 100 94

基线状态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 7.3 /10 ×20 — 10.0 /10 ×20 10.0 /10 ×20 7.3 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 — 4.0 /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

缓存连续性 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

原生 msg-ID 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 — 0.0 /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

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 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

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

延迟基线 10.0 /10 ×5 — 9.6 /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

stop_reason 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

流式投递未计入 — 未计入未计入未计入

综合结论

与基线相似度 93 v3.21.0 执行失败，不评分 79 v3.21.0 基准（参考） 94 v3.21.0

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 自由身份 identity_free_clean
5 停止原因 stop_reason_present
6 系统提示纯净 system_prompt_clean
7 Service Tier service_tier_present
8 缓存命中比 cache_hit_ratio_match
9 缓存 TTL 一致 cache_ttl_consistency
10 推理地区 inference_geo_present
11 缓存连续 cache_continuity_intra
12 知识召回 knowledge_recall_match
13 延迟基线 latency_baseline_match
14 原生 msg-ID anthropic_msg_id_format
15 Sliding 缓存 cache_sliding_correctness
16 结构化身份 identity_structured_match
17 信封自报一致 envelope_self_report_match
18 中文遵循 instruction_following_lang
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 4256 ms 2619 ms 5606 ms 5110 ms 6077 ms

输出速度 76.7 tok/s - burst 77.8 tok/s 86.9 tok/s

完成步骤 6/6 成功 5/6 成功 · 1 失败 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 ⓘ

$0.1972

$0.1835 (部分)

$0.0952

$0.0675

$0.2063

缓存读取占比 ⓘ 70% 71% (部分) 81% 98% 70%

揭晓

通道

服务商 PackyCode ↗

通道cc-Zeus

服务商 SSSAiCode ↗

通道O-Max

服务商 luminai ↗

通道M-Kiro/Antigravity

官方基准

服务商 RunAPI ↗

通道O-Max

返回列表