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

2026/06/25 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

DawAPI

官方基准

WorldBase.ai

M-

FastCode

0-0

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 01:34:07 ~ 01:34:22 (11.8s) 01:34:13 ~ 01:34:18 (4.9s) 01:34:05 ~ 01:34:20 (12.1s) 01:34:09 ~ 01:34:24 (10.7s) 01:34:11 ~ 01:34:21 (7.9s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 10,258 ms 3,464 ms 10,407 ms 6,659 ms 3,998 ms

首文本(SSE delta) 10,299 ms 3,466 ms 10,736 ms 6,661 ms 3,999 ms

总耗时 10,680 ms 3,646 ms 10,736 ms 8,831 ms 4,392 ms

输入 3 3 14 3 3

输出 5 5 2 5 5

缓存创建

32,316

5m: 0 / 1h: 32,316

32,318

5m: 0 / 1h: 32,318

32,611

5m: 0 / 1h: 32,611

32,318

5m: 0 / 1h: 32,318

32,805

5m: 0 / 1h: 32,805

缓存读取 0 0 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 01:36:22 ~ 01:36:32 (5.5s) 01:36:10 ~ 01:36:15 (3.8s) 01:37:07 ~ 01:37:17 (6.5s) 01:35:46 ~ 01:35:56 (8.2s) 01:36:04 ~ 01:36:14 (5.8s)

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。请只输出以下 3 行内容（label 小写，值小写）：
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) 4,321 ms 2,371 ms 4,842 ms 6,593 ms 3,655 ms

首文本(SSE delta) 4,322 ms 2,373 ms 4,856 ms 6,595 ms 3,656 ms

总耗时 4,695 ms 2,759 ms 5,325 ms 6,987 ms 3,980 ms

输入 3 3 220 3 3

输出 22 22 16 22 22

缓存创建

178

5m: 0 / 1h: 178

178

5m: 0 / 1h: 178

32,626

5m: 0 / 1h: 32,626

178

5m: 0 / 1h: 178

7,621

5m: 0 / 1h: 7,621

缓存读取 32,316 32,318 0 32,318 25,362

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 01:39:56 ~ 01:40:21 (21.8s) 01:38:47 ~ 01:38:52 (3.9s) 01:38:29 ~ 01:38:39 (6.3s) 01:39:58 ~ 01:40:13 (13.6s) 01:38:25 ~ 01:38:35 (5.3s)

提示词

查看本步 Prompt

你掌握的世界知识截止于何时？YYYY-MM 格式作答，不解释。

HTTP 头(TTFB) 20,655 ms 2,389 ms 4,720 ms 9,618 ms 3,694 ms

首文本(SSE delta) 20,656 ms 2,604 ms 4,979 ms 9,619 ms 3,695 ms

总耗时 20,986 ms 2,969 ms 4,998 ms 12,205 ms 3,897 ms

输入 1 3 41 3 3

输出 9 8 3 8 8

缓存创建

31,950

5m: 0 / 1h: 31,950

5m: 0 / 1h: 55

241

5m: 0 / 1h: 241

5m: 0 / 1h: 38

7,676

5m: 0 / 1h: 7,676

缓存读取 0 32,496 32,626 32,496 25,362

响应

2025-08

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 01:43:01 ~ 01:43:21 (15.9s) 01:41:51 ~ 01:41:56 (4.4s) 01:40:21 ~ 01:40:31 (5.2s) 01:43:43 ~ 01:44:03 (17.7s) 01:42:13 ~ 01:42:23 (7.1s)

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) 3,462 ms 2,376 ms 3,281 ms 12,714 ms 3,369 ms

首文本(SSE delta) 3,463 ms 2,378 ms 3,282 ms 12,715 ms 3,370 ms

总耗时 4,451 ms 3,349 ms 3,860 ms 16,388 ms 5,535 ms

输入 1 3 31,446 3 3

输出 48 66 39 64 66

缓存创建

1,803

5m: 0 / 1h: 1,803

5m: 0 / 1h: 43

1,510

5m: 0 / 1h: 1,510

5m: 0 / 1h: 43

7,719

5m: 0 / 1h: 7,719

缓存读取 30,417 32,551 0 32,534 25,362

输出速度 48.6 tok/s 68.0 tok/s 67.5 tok/s 17.4 tok/s 30.5 tok/s

响应

我叫 Claude Code，是 Anthropic 开发的 AI 编程助手，基于 Claude 模型构建，运行在命令行终端中。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助完成写作、分析、编程、问答等多种任务，致力于提供安全、有益、诚实的服务。

我是 Claude，由 Anthropic 开发的 AI 助手。我的当前版本是 claude-sonnet-4-6，擅长代码、分析、写作等多种任务。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助编写代码、分析问题、回答问题等多种任务。当前版本为 claude-sonnet-4-6。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助您完成写作、分析、编程、问答等各类任务，致力于提供安全、准确、有益的服务。

步骤 5: knowledge_recall text

时间 01:45:04 ~ 01:45:14 (8.4s) 01:44:03 ~ 01:44:08 (4.6s) 01:41:57 ~ 01:42:07 (8.4s) 01:47:57 ~ 01:48:12 (12.4s) 01:44:12 ~ 01:44:22 (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) 2,147 ms 1,694 ms 5,103 ms 5,967 ms 2,623 ms

首文本(SSE delta) 2,149 ms 1,695 ms 5,204 ms 5,969 ms 2,624 ms

总耗时 4,198 ms 3,443 ms 7,389 ms 11,035 ms 5,084 ms

输入 1 3 753 3 3

输出 211 205 159 204 205

缓存创建

2,905

5m: 0 / 1h: 2,905

8,049

5m: 0 / 1h: 8,049

161

5m: 0 / 1h: 161

8,030

5m: 0 / 1h: 8,030

8,465

5m: 0 / 1h: 8,465

缓存读取 6,164 0 7,776 0 0

输出速度 103.0 tok/s 117.3 tok/s 72.8 tok/s 40.3 tok/s 83.3 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 01:47:33 ~ 01:47:53 (17.8s) 01:46:47 ~ 01:47:07 (18.4s) 01:44:23 ~ 01:44:33 (5.2s) 01:51:50 ~ 01:52:10 (17.7s) 01:46:01 ~ 01:46:21 (15.5s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=311921826759663787547315051106312575579952012736602406183970503534416246

HTTP 头(TTFB) 4,902 ms 2,270 ms 3,609 ms 1,626 ms 2,449 ms

首文本(SSE delta) 14,955 ms 2,272 ms 3,610 ms 1,655 ms 2,864 ms

总耗时 17,093 ms 17,445 ms 4,187 ms 16,482 ms 14,023 ms

输入 3 164 1 164 581

输出 1,197 1,649 18 1,003 952

缓存创建

36,382

5m: 0 / 1h: 36,382

5m: 0 / 1h: 0

244

5m: 348 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 156 0 0

输出速度 burst 108.7 tok/s - 67.6 tok/s 85.3 tok/s

响应

Going through `s = "311921826759663787547315051106312575579952012736602406183970503534416246"` digit by digit:

**Length:** 72 digits

**Counting 0s** (positions 25, 29, 43, 50, 53, 60, 62): **7**

**

**Counting the string:**

`s = 311921826759663787547315051106312575579952012736602406183970503534416246`

**Splitting into groups of 10:**
```
3119218267  (1–10)
5966378754  (11–20)
7315051106  (21–30

{"n":75,"c":[7,6,6]}

```json
{"n":72,"c":[7,8,5]}
```

```json
{"n":72,"c":[7,8,5]}
```

总计

总输入(含缓存) 174,443 138,187 140,426 138,134 140,968

总输出 1,492 1,955 237 1,306 1,258

总缓存创建

105,534

5m: 0 / 1h: 105,534

40,643

5m: 0 / 1h: 40,643

67,393

5m: 348 / 1h: 67,149

40,607

5m: 0 / 1h: 40,607

64,286

5m: 0 / 1h: 64,286

总缓存读取 68,897 97,365 40,558 97,348 76,086

输出速度 85.3 tok/s 107.3 tok/s 71.7 tok/s 53.9 tok/s 77.5 tok/s

总执行时间 1m 21.3s 40.0s 43.6s 1m 20.4s 47.9s

总墙钟时间

13m 46.1s

含步骤间等待

12m 54.0s

含步骤间等待

10m 28.0s

含步骤间等待

18m 1.2s

含步骤间等待

12m 10.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.4388

Anthropic: Claude Sonnet 4.6

$0.2115

Anthropic: Claude Sonnet 4.6

$0.3658

Anthropic: Claude Sonnet 4.6

$0.2016

Anthropic: Claude Sonnet 4.6

$0.2846

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%) direct (100%) direct (100%) direct (100%) direct (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_014zrPWFE9keAHbc539JQ4Qz msg_017AtdBAKa436cBy5MAidCFq msg_01cNVfNR2bQfrPwX4pPGBFUj msg_015a5pxkmMfwMq68qbQjB3u8 msg_01NFGSiMEWYdByeUGohLZGnK

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	server: cdn x-new-api-version: v1.0.0-rc.14 x-oneapi-request-id: 202606241747339332442468268d9d6XkmNO133	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_011CcNW9pepQcdZonHdS68yg
M-	cf-ray: a10d8a4fcdf4773c-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	becd4e55-53e0-4d5a-bc1a-ee45a90193de a10d8a4fcdf4773c-AMS
cc	server: nginx	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	—
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	1efe9113a8b16e300fcef3a930b3ada6

协议指纹评分（机器自动）

总分 / 100 84 100 73 94 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已对照基线

缓存命中比 6.9 /10 ×20 10.0 /10 ×20 0.0 /10 ×20 10.0 /10 ×20 8.9 /10 ×20

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

sliding 5m 7.7 /10 ×13 10.0 /10 ×13 5.3 /10 ×13 10.0 /10 ×13 10.0 /10 ×13

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

系统提示纯净 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 5.0 /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 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 10.0 /10 ×5

延迟基线 5.0 /10 ×5 10.0 /10 ×5 9.2 /10 ×5 0.0 /10 ×5 10.0 /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 10.0 /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 84 v3.24.1 基准（参考） 73 v3.24.1 94 v3.24.1 94 v3.24.1

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 10351 ms 5602 ms 6083 ms 11988 ms 6152 ms

输出速度 85.3 tok/s 107.3 tok/s 71.7 tok/s 53.9 tok/s 77.5 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 ⓘ

$0.4388

$0.2115

$0.3658

$0.2016

$0.2846

缓存读取占比 ⓘ 39% 70% 29% 70% 54%

揭晓

通道

服务商 DawAPI ↗

通道cc

官方基准

服务商 WorldBase.ai ↗

通道M-

服务商 FastCode ↗

通道cc

服务商 0-0 ↗

通道O-Max

返回列表