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

2026/06/20 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

官方基准

SAIAi

O-Max

SSSAiCode

O-Max-L2

Novaris

O-Max

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 12:11:47 ~ 12:11:52 (4.0s) 12:11:43 ~ 12:11:48 (4.3s) 12:11:39 ~ 12:11:44 (3.4s) 12:11:45 ~ 12:11:55 (4.9s) 12:11:41 ~ 12:11:51 (5.1s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,730 ms 3,158 ms 2,223 ms 3,815 ms 3,362 ms

首文本(SSE delta) 2,731 ms 3,161 ms 2,306 ms 3,816 ms 3,363 ms

总耗时 2,875 ms 3,265 ms 2,368 ms 4,012 ms 3,414 ms

输入 3 3 3 3 3

输出 5 5 5 5 5

缓存创建

5m: 0 / 1h: 0

32,318

5m: 0 / 1h: 32,318

31,990

5m: 0 / 1h: 31,990

25,697

5m: 25,697 / 1h: 0

32,319

5m: 0 / 1h: 32,319

缓存读取 32,318 0 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 12:13:53 ~ 12:13:58 (3.5s) 12:14:24 ~ 12:14:29 (4.1s) 12:14:48 ~ 12:14:53 (3.0s) 12:14:29 ~ 12:14:34 (3.4s) 12:14:43 ~ 12:14:53 (5.0s)

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) 2,091 ms 2,483 ms 1,940 ms 2,032 ms 2,784 ms

首文本(SSE delta) 2,093 ms 2,486 ms 2,208 ms 2,123 ms 2,786 ms

总耗时 2,451 ms 3,117 ms 2,271 ms 2,560 ms 3,228 ms

输入 3 3 3 3 3

输出 22 22 22 22 22

缓存创建

189

5m: 0 / 1h: 189

5m: 0 / 1h: 0

189

5m: 0 / 1h: 189

189

5m: 189 / 1h: 0

189

5m: 0 / 1h: 189

缓存读取 32,318 32,507 31,990 25,697 32,319

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-4-5

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

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

步骤 3: cutoff text

时间 12:16:11 ~ 12:16:21 (5.2s) 12:18:28 ~ 12:18:33 (3.3s) 12:16:55 ~ 12:17:00 (2.9s) 12:16:58 ~ 12:17:03 (3.4s) 12:16:26 ~ 12:16:31 (4.4s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 3,996 ms 2,007 ms 1,932 ms 2,274 ms 2,811 ms

首文本(SSE delta) 4,115 ms 2,118 ms 2,086 ms 2,560 ms 2,812 ms

总耗时 4,189 ms 2,196 ms 2,152 ms 2,627 ms 3,125 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: 57

5m: 57 / 1h: 0

5m: 0 / 1h: 38

缓存读取 32,507 32,507 32,179 25,886 32,508

响应

2025-08

2025-08

2025-04

2025-04

2025-08

步骤 4: identity_free text

时间 12:20:14 ~ 12:20:19 (4.7s) 12:21:13 ~ 12:21:18 (4.4s) 12:19:03 ~ 12:19:08 (3.2s) 12:18:07 ~ 12:18:12 (3.5s) 12:19:21 ~ 12:19:31 (5.0s)

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) 2,815 ms 2,431 ms 1,985 ms 2,042 ms 2,584 ms

首文本(SSE delta) 3,220 ms 2,739 ms 2,272 ms 2,043 ms 2,586 ms

总耗时 3,666 ms 3,365 ms 2,331 ms 2,818 ms 3,612 ms

输入 3 3 3 3 3

输出 46 57 26 54 44

缓存创建

5m: 0 / 1h: 41

5m: 0 / 1h: 43

5m: 0 / 1h: 39

5m: 43 / 1h: 0

5m: 0 / 1h: 39

缓存读取 32,545 32,564 32,236 25,943 32,546

输出速度 103.1 tok/s 91.1 tok/s - 69.7 tok/s 42.9 tok/s

响应

我叫 Claude，是由 Anthropic 开发的 AI 助手，当前使用的模型版本为 claude-sonnet-4-6。

我是 Claude，由 Anthropic 开发的 AI 助手。当前版本为 Claude Sonnet 4.6，擅长对话、分析、编程和写作等任务。

我是 Claude，由 Anthropic 公司打造的 AI 助手。

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，由 Anthropic 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 12:21:31 ~ 12:21:46 (12.0s) 12:23:54 ~ 12:23:59 (4.8s) 12:21:51 ~ 12:22:01 (5.4s) 12:19:34 ~ 12:19:39 (4.3s) 12:22:30 ~ 12:22:40 (6.3s)

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) 1,374 ms 1,852 ms 1,724 ms 1,555 ms 2,371 ms

首文本(SSE delta) 1,753 ms 1,854 ms 3,909 ms 1,645 ms 2,394 ms

总耗时 11,005 ms 3,807 ms 4,678 ms 3,499 ms 4,832 ms

输入 3 3 3 3 3

输出 794 205 198 198 204

缓存创建

8,021

5m: 0 / 1h: 8,021

2,335

5m: 0 / 1h: 2,335

7,690

5m: 0 / 1h: 7,690

2,348

5m: 2,348 / 1h: 0

8,018

5m: 0 / 1h: 8,018

缓存读取 0 5,718 0 0 0

输出速度 85.8 tok/s 105.0 tok/s burst 106.8 tok/s 83.7 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 12:25:33 ~ 12:25:53 (19.2s) 12:26:11 ~ 12:26:36 (20.6s) 12:25:24 ~ 12:25:29 (1.1s) 12:22:32 ~ 12:22:52 (17.6s) 12:24:51 ~ 12:25:11 (17.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=504720634905832141190510249835561655690998794887993046667026942116015487

HTTP 头(TTFB) 1,166 ms 877 ms - 1,972 ms 1,380 ms

首文本(SSE delta) 1,168 ms 879 ms - 1,973 ms 1,383 ms

总耗时 18,240 ms 19,531 ms - 16,713 ms 16,437 ms

输入 164 164 - 154 164

输出 1,720 1,881 - 1,487 1,120

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 - 0 0

输出速度 100.7 tok/s 100.8 tok/s - 100.9 tok/s 74.4 tok/s

响应

```
s = "504720634905832141190510249835561655690998794887993046667026942116015487"
```

**Mapping each character (1-indexed):**

| # | digit | | # | digit | | # | digit | | # | digit |
|---|-------|-|

## Counting digits in s

**s** = `504720634905832141190510249835561655690998794887993046667026942116015487`

### Step 1 — Length (groups of 10)

```
5047206349  → 10
0583214119  → 20
0510249835  → 30

失败

评测失败

## Counting digits in s

**s =** `504720634905832141190510249835561655690998794887993046667026942116015487`

### Step 1 — Length (n)

Breaking into groups of 10:

```
5047206349  →  10
0583214119  →

```json
{"n":72,"c":[9,5,10]}
```

**Verification:**

**String:** `504720634905832141190510249835561655690998794887993046667026942116015487`

**Length (n=72):** Counted in groups of 10:
`5047206349` |

总计

总输入(含缓存)

138,156

138,228

136,385

仅前 6 步

106,029

138,155

总输出

2,595

2,178

259

仅前 6 步

1,774

1,403

总缓存创建

8,289

5m: 0 / 1h: 8,289

34,753

5m: 0 / 1h: 34,753

39,965

5m: 0 / 1h: 39,965

仅前 6 步

28,334

5m: 28,334 / 1h: 0

40,603

5m: 0 / 1h: 40,603

总缓存读取

129,688

103,296

96,405

仅前 6 步

77,526

97,373

输出速度

95.6 tok/s

100.9 tok/s

burst

仅前 6 步

100.1 tok/s

73.9 tok/s

总执行时间

48.5s

41.5s

19.0s

仅前 6 步

37.2s

43.7s

总墙钟时间

14m 6.0s

含步骤间等待

14m 53.0s

含步骤间等待

13m 49.9s

含步骤间等待

仅前 6 步

11m 6.8s

含步骤间等待

13m 30.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1095

Anthropic: Claude Sonnet 4.6

$0.1945

Anthropic: Claude Sonnet 4.6

$0.1827

Anthropic: Claude Sonnet 4.6 (部分)

$0.1168

Anthropic: Claude Sonnet 4.6

$0.2031

Anthropic: Claude Sonnet 4.6

通道指纹

来源步骤 digit_count digit_count knowledge_recall digit_count digit_count

平台 direct (100%) direct (100%) direct (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 - - - - -

ID 格式 msg_01NjDEWx1NxB5hT6S2mMqHUH msg_012LSVDTgYU3ikLacze75Cno msg_01Dt7TQKvx4MLb6cYmTWVesi msg_01QC2e7fujQ9soUVrQUq5UHe msg_016RsfuiiXeQAdcuYU6a9wmF

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_011CcDspQfvTXxuydeUTcCEt
O-Max	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_011CcDssDChCQaskv64iMZy1
O-Max-L2	cf-ray: a0e7fd39b972b8d4-LAX server: Photon-Edge set-cookie: _cfuvid=9hnCmtRFXq3QxD.ks9iW99JeW4Y_sRSl4MBLyBPqv.I-1781929312.2753334-1.0.1.1-1FquOZmLpL.Ji5dUeCJ5nyj2oSIYnH3fq0192h9OXLI; 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	1781929312226-kbag4jbm1m1a4 req_011CcDsY2vn6GKFKG1EcbJdj a0e7fd39b972b8d4-LAX
O-Max	server: nginx x-oneapi-request-id: 20260620042233593424428268d9d6SqCJTWny	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	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606200424528643183758268d9d6C9AGnz3y	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 97 100 77 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

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

缓存 TTL 一致 10.0 /10 ×15 10.0 /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

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 10.0 /10 ×8 — 0.0 /10 ×8 10.0 /10 ×8

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

知识截止 10.0 /10 ×7 10.0 /10 ×7 — 5.0 /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 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

流式投递 0.0 /10 ×5 10.0 /10 ×5 — 10.0 /10 ×5 10.0 /10 ×5

Req-ID 透传 10.0 /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

SDK 一致 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） 100 v3.24.0 执行失败，不评分 77 v3.24.0 97 v3.24.0

雷达图例

共 22 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 Thinking 存在 thinking_present
5 流式投递 buffer_dump_match
6 自由身份 identity_free_clean
7 停止原因 stop_reason_present
8 系统提示纯净 system_prompt_clean
9 Service Tier service_tier_present
10 缓存命中比 cache_hit_ratio_match
11 缓存 TTL 一致 cache_ttl_consistency
12 推理地区 inference_geo_present
13 缓存连续 cache_continuity_intra
14 知识召回 knowledge_recall_match
15 延迟基线 latency_baseline_match
16 原生 msg-ID anthropic_msg_id_format
17 Sliding 缓存 cache_sliding_correctness
18 结构化身份 identity_structured_match
19 信封自报一致 envelope_self_report_match
20 中文遵循 instruction_following_lang
21 档位思考量 tier_thinking_volume_match
22 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 7071 ms 5880 ms 2760 ms 5372 ms 5775 ms

输出速度 95.6 tok/s 100.9 tok/s burst 100.1 tok/s 73.9 tok/s

完成步骤 6/6 成功 6/6 成功 5/6 成功 · 1 失败 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 ⓘ

$0.1095

$0.1945

$0.1827 (部分)

$0.1168

$0.2031

缓存读取占比 ⓘ 94% 75% 71% (部分) 73% 70%

揭晓

通道

官方基准

服务商 SAIAi ↗

通道O-Max

服务商 SSSAiCode ↗

通道O-Max-L2

服务商 Novaris ↗

通道O-Max

服务商 ClaudeAPI ↗

通道O-Bdrk

返回列表