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

2026/06/28 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

YunWu

O-Api

SudoCode

M-Max

LingxiCode

M-

0-0

O-Max

官方基准

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 02:28:15 ~ 02:28:20 (3.7s) 02:28:17 ~ 02:28:27 (5.0s) 02:28:21 ~ 02:28:31 (5.0s) 02:28:19 ~ 02:28:29 (6.8s) 02:28:23 ~ 02:28:28 (4.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,523 ms 3,205 ms 3,487 ms 5,094 ms 2,631 ms

首文本(SSE delta) 2,524 ms 3,217 ms 3,488 ms 5,096 ms 2,633 ms

总耗时 2,541 ms 3,884 ms 3,862 ms 5,131 ms 2,959 ms

输入 8,598 3 3 3 3

输出 2 5 5 5 5

缓存创建

32,318

5m: 0 / 1h: 32,318

32,317

5m: 0 / 1h: 32,317

32,821

5m: 0 / 1h: 32,821

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 02:29:28 ~ 02:29:33 (4.2s) 02:31:15 ~ 02:31:25 (5.3s) 02:32:01 ~ 02:32:11 (6.0s) 02:31:02 ~ 02:31:12 (6.1s) 02:30:17 ~ 02:30:22 (3.1s)

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) 2,857 ms 3,214 ms 4,635 ms 4,480 ms 1,663 ms

首文本(SSE delta) 2,858 ms 3,215 ms 4,637 ms 4,481 ms 1,665 ms

总耗时 2,960 ms 4,190 ms 5,212 ms 4,807 ms 2,178 ms

输入 8,797 3 3 3 3

输出 21 22 22 22 22

缓存创建

192

5m: 0 / 1h: 192

192

5m: 0 / 1h: 192

32,997

5m: 0 / 1h: 32,997

192

5m: 0 / 1h: 192

缓存读取 0 32,318 32,317 0 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 02:31:05 ~ 02:31:15 (5.0s) 02:34:55 ~ 02:35:00 (4.6s) 02:36:09 ~ 02:36:14 (4.2s) 02:32:59 ~ 02:33:09 (7.2s) 02:32:50 ~ 02:32:55 (3.8s)

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) 3,637 ms 3,394 ms 2,851 ms 5,607 ms 2,564 ms

首文本(SSE delta) 3,666 ms 3,395 ms 2,852 ms 5,901 ms 2,673 ms

总耗时 3,721 ms 3,643 ms 3,274 ms 5,914 ms 2,745 ms

输入 8,854 3 3 3 3

输出 5 8 8 8 8

缓存创建

5m: 0 / 1h: 57

7,673

5m: 0 / 1h: 7,673

5m: 0 / 1h: 57

缓存读取 0 32,510 32,509 25,362 32,510

响应

2025-05

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 02:34:47 ~ 02:34:52 (4.0s) 02:37:08 ~ 02:37:18 (6.6s) 02:38:47 ~ 02:38:57 (5.3s) 02:34:35 ~ 02:34:45 (5.6s) 02:36:47 ~ 02:36:53 (4.0s)

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) 2,693 ms 4,534 ms 3,228 ms 3,263 ms 2,431 ms

首文本(SSE delta) 2,695 ms 4,535 ms 3,229 ms 3,641 ms 2,789 ms

总耗时 2,805 ms 5,519 ms 4,409 ms 4,110 ms 2,907 ms

输入 8,891 3 3 3 3

输出 14 55 63 55 39

缓存创建

5m: 0 / 1h: 35

5m: 0 / 1h: 41

7,717

5m: 0 / 1h: 7,717

5m: 0 / 1h: 35

缓存读取 0 32,567 32,566 25,362 32,567

输出速度 - 55.9 tok/s 53.4 tok/s 117.3 tok/s -

响应

我叫Claude，属于Claude产品，由Anthropic公司开发。

我是 Claude，由 Anthropic 开发的 AI 助手。我基于大型语言模型构建，旨在提供安全、有益的对话和任务协助。

我是 Claude，由 Anthropic 开发的 AI 助手。当前版本为 Claude Sonnet 4.6，是 Anthropic 推出的 Claude 系列大语言模型产品之一。

我是 Claude，由 Anthropic 开发的 AI 助手。我擅长编程、写作、分析和对话，致力于提供安全、有用的帮助。

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。

步骤 5: knowledge_recall text

时间 02:37:15 ~ 02:37:20 (4.6s) 02:40:29 ~ 02:40:39 (5.3s) 02:41:02 ~ 02:41:12 (5.9s) 02:36:33 ~ 02:36:48 (13.9s) 02:39:59 ~ 02:40:09 (5.0s)

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) 2,607 ms 1,899 ms 2,398 ms 2,242 ms 1,565 ms

首文本(SSE delta) 2,609 ms 1,900 ms 2,399 ms 2,665 ms 1,567 ms

总耗时 3,468 ms 4,267 ms 5,043 ms 12,471 ms 4,007 ms

输入 9,500 3 3 3 3

输出 160 205 205 789 204

缓存创建

8,046

5m: 0 / 1h: 8,046

8,059

5m: 0 / 1h: 8,059

8,452

5m: 0 / 1h: 8,452

8,030

5m: 0 / 1h: 8,030

缓存读取 0 0 0 0 0

输出速度 burst 86.6 tok/s 77.5 tok/s 80.5 tok/s 83.6 tok/s

响应

{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Past Lives","ucl_2024":"Manchester City","nba_2024":"Denver Nuggets","copa_america_2024":"Argentina","euro2024":"Germany","

{"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 02:38:54 ~ 02:38:59 (3.3s) 02:41:43 ~ 02:41:59 (13.5s) 02:44:29 ~ 02:44:49 (15.1s) 02:38:38 ~ 02:39:08 (26.1s) 02:41:28 ~ 02:41:43 (13.6s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=252340603323134819647805865767584695638386058641104145227278316099165197

HTTP 头(TTFB) 1,885 ms 1,316 ms 2,051 ms 2,473 ms 1,195 ms

首文本(SSE delta) 1,886 ms 1,341 ms 2,473 ms 2,847 ms 1,196 ms

总耗时 2,001 ms 12,174 ms 14,394 ms 24,847 ms 12,666 ms

输入 166 164 164 3 164

输出 18 900 991 1,269 1,034

缓存创建

5m: 0 / 1h: 0

1,042

5m: 1,042 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 - 83.1 tok/s 83.1 tok/s 57.7 tok/s 90.1 tok/s

响应

```json
{"n":69,"c":[6,6,5]}
```

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

**Verification:**

| digit | positions (1-indexed) | count |
|---|---|---|
| **0** | 6, 8, 23, 43, 50, 64 | **6** |
| **7** | 21, 28, 30, 57, 59, 72 | **6** |
| **9**

**Parsing `s` digit by digit (length = 72):**

```
2 5 2 3 4 0 6 0 3 3  (10)
2 3 1 3 4 8 1 9 6 4  (10)
7 8 0 5 8 6 5 7 6 7  (10)
5 8 4 6 9 5 6 3 8 3  (10)
8 6 0 5 8 6 4 1 1 0  (10)
4 1 4 5 2 2 7 2 7 8

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

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

**Verification breakdown for `s = 252340603323134819647805865767584695638386058641104145227278316099165197`:**

| Digit | Positions (1-indexed) | Count |
|---|---|---

总计

总输入(含缓存) 44,806 138,222 138,237 141,444 138,206

总输出 220 1,195 1,294 2,148 1,312

总缓存创建

40,648

5m: 0 / 1h: 40,648

40,666

5m: 0 / 1h: 40,666

90,702

5m: 1,042 / 1h: 89,660

40,632

5m: 0 / 1h: 40,632

总缓存读取 0 97,395 97,392 50,724 97,395

输出速度 burst 81.8 tok/s 80.0 tok/s 65.5 tok/s 89.0 tok/s

总执行时间 24.8s 40.3s 41.5s 1m 5.8s 33.8s

总墙钟时间

10m 43.9s

含步骤间等待

13m 41.6s

含步骤间等待

16m 28.3s

含步骤间等待

10m 49.0s

含步骤间等待

13m 20.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1377

Anthropic: Claude Sonnet 4.6

$0.2001

Anthropic: Claude Sonnet 4.6

$0.2017

Anthropic: Claude Sonnet 4.6

$0.3862

Anthropic: Claude Sonnet 4.6

$0.2018

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 unknown (0%) one-api (100%) one-api (100%) direct (100%) direct (100%)

上游 - Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%)

CDN - - cloudflare - -

ID 格式自定义 msg_01UipC8YfPXLaEFJZadxnbva msg_016t6YjwVj61ko7g1rYHz7jd msg_0152JbsRsLWKJmviFW8XwMEK msg_01CYKmQGRskifwYwSwGsmioM

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-Api	server: nginx	id_format: non_standard has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	body:id:non_standardbody:usage.cache_fields	—
M-Max	server: nginx x-new-api-version: 0b8be5cf x-oneapi-request-id: 20260628024145412763019S04pq8mt	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	—
M-	cf-ray: a1269a79ec21e041-NRT server: cloudflare set-cookie: server_name_session=6268b44f10666c8b08965a0a61ec74a7; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 20260627184431383706488hc2r6I1	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-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a1269a79ec21e041-NRT
O-Max	via: 1.1 Caddy server: uvicorn, nginx/1.24.0 (Ubuntu)	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:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	d4d9c2b997dfe5207c2673efb84acc7d
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_011CcUFkRRaQoXJNexSt3UDS

协议指纹评分（机器自动）

总分 / 100 37 97 98 78 100

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 0.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 0.4 /10 ×20 10.0 /10 ×20

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

缓存连续性 5.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

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

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

系统提示纯净 0.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 5.0 /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 5.0 /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 0.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

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

延迟基线 10.0 /10 ×5 8.7 /10 ×5 10.0 /10 ×5 6.7 /10 ×5 10.0 /10 ×5

流式投递 3.1 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 0.0 /10 ×5 10.0 /10 ×5

Req-ID 透传 0.0 /10 ×4 0.0 /10 ×4 0.0 /10 ×4 0.0 /10 ×4 10.0 /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 37 v3.25.0 97 v3.25.0 98 v3.25.0 78 v3.25.0 基准（参考）

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 2916 ms 5613 ms 6032 ms 9547 ms 4577 ms

输出速度 burst 81.8 tok/s 80.0 tok/s 65.5 tok/s 89.0 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 ⓘ

$0.1377

$0.2001

$0.2017

$0.3862

$0.2018

缓存读取占比 ⓘ 0% 70% 70% 36% 70%

揭晓

通道

服务商 YunWu ↗

通道O-Api

服务商 SudoCode ↗

通道M-Max

服务商 LingxiCode ↗

通道M-

服务商 0-0 ↗

通道O-Max

官方基准

返回列表