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-haiku-4-5-20251001 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

ByteCatCode

O-Max

SudoCode

M-Max

AnyAIGC

M-

官方基准

duckcoding

cc专用

基本信息

服务名 Claude Code Claude Code Claude Code Claude Code Claude Code

模型 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001

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 23:39:22 ~ 23:39:32 (5.3s) 23:39:28 ~ 23:39:33 (4.3s) 23:39:26 ~ 23:39:36 (5.4s) 23:39:30 ~ 23:39:40 (5.4s) 23:39:24 ~ 23:39:39 (12.4s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,886 ms 3,003 ms 3,959 ms 3,919 ms 10,784 ms

首文本(SSE delta) - 3,004 ms 3,961 ms 3,922 ms 10,786 ms

总耗时 3,756 ms 3,287 ms 3,961 ms 4,280 ms 11,124 ms

输入 10 10 8 10 10

输出 52 41 2 49 44

缓存创建

33,169

5m: 0 / 1h: 33,169

33,122

5m: 0 / 1h: 33,122

7,138

5m: 0 / 1h: 7,138

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 29,572 0 0

输出速度 59.8 tok/s 144.9 tok/s - 136.9 tok/s 130.2 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 23:43:23 ~ 23:43:33 (6.9s) 23:41:34 ~ 23:41:44 (5.6s) 23:42:34 ~ 23:42:44 (5.3s) 23:41:24 ~ 23:41:34 (6.7s) 23:42:01 ~ 23:42:21 (19.2s)

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) 2,806 ms 2,916 ms 3,994 ms 3,622 ms 16,133 ms

首文本(SSE delta) - 2,917 ms 4,024 ms 3,624 ms 16,134 ms

总耗时 5,654 ms 4,572 ms 4,035 ms 5,727 ms 18,220 ms

输入 10 10 8 10 10

输出 192 226 22 263 231

缓存创建

189

5m: 0 / 1h: 189

33,311

5m: 0 / 1h: 33,311

7,154

5m: 0 / 1h: 7,154

240

5m: 0 / 1h: 240

235

5m: 0 / 1h: 235

缓存读取 33,169 0 28,639 33,122 33,122

输出速度 67.4 tok/s 136.6 tok/s - 125.1 tok/s 110.7 tok/s

响应

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

步骤 3: cutoff text

时间 23:47:03 ~ 23:47:13 (6.0s) 23:44:15 ~ 23:44:20 (4.7s) 23:46:04 ~ 23:46:14 (5.9s) 23:44:01 ~ 23:44:11 (8.0s) 23:43:51 ~ 23:44:16 (21.6s)

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) 3,643 ms 2,957 ms 4,368 ms 6,393 ms 20,117 ms

首文本(SSE delta) - 2,959 ms 4,369 ms 6,394 ms 20,118 ms

总耗时 4,587 ms 3,608 ms 4,713 ms 6,908 ms 20,611 ms

输入 10 10 8 10 10

输出 75 121 4 91 74

缓存创建

5m: 0 / 1h: 59

5m: 0 / 1h: 61

7,069

5m: 0 / 1h: 7,069

305

5m: 0 / 1h: 305

271

5m: 0 / 1h: 271

缓存读取 33,358 33,311 28,625 33,362 33,357

输出速度 79.4 tok/s 186.4 tok/s - 177.0 tok/s 150.1 tok/s

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 23:48:22 ~ 23:48:32 (7.1s) 23:48:16 ~ 23:48:26 (7.2s) 23:49:54 ~ 23:50:04 (5.1s) 23:45:44 ~ 23:45:55 (7.4s) 23:47:33 ~ 23:47:54 (19.3s)

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) 3,918 ms 4,746 ms 3,325 ms 3,857 ms 11,755 ms

首文本(SSE delta) - 4,747 ms 3,482 ms 3,859 ms 11,775 ms

总耗时 5,662 ms 6,179 ms 3,785 ms 6,331 ms 18,315 ms

输入 10 10 8 10 10

输出 175 211 55 442 809

缓存创建

5m: 0 / 1h: 41

33,413

5m: 0 / 1h: 33,413

7,003

5m: 0 / 1h: 7,003

133

5m: 0 / 1h: 133

116

5m: 0 / 1h: 116

缓存读取 33,417 0 28,720 33,667 33,628

输出速度 100.3 tok/s 147.3 tok/s 181.5 tok/s 178.8 tok/s 123.7 tok/s

响应

我叫Claude Haiku，是Anthropic公司开发的Claude大语言模型系列中的轻量级版本。属于Claude产品，由Anthropic公司开发。

我是Claude Haiku 4.5模型，属于Claude产品系列，由Anthropic公司开发。

我是 Claude Code，由 Anthropic 打造的 AI 编程助手。基于 Claude 模型，帮助开发者完成代码编写、调试、测试等软件工程任务。

我是Claude，由Anthropic公司开发的AI助手。我可以帮助用户进行软件工程任务、代码编写、问题解答和信息研究等工作。

我是Claude，由Anthropic开发的AI助手。Haiku模型，可帮助编程、分析和创意工作。

步骤 5: knowledge_recall text

时间 23:52:25 ~ 23:52:55 (28.6s) 23:50:28 ~ 23:50:48 (15.0s) 23:53:09 ~ 23:53:19 (5.2s) 23:49:47 ~ 23:50:07 (17.0s) 23:51:53 ~ 23:52:18 (22.3s)

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) 2,202 ms 2,107 ms 2,378 ms 2,389 ms 9,455 ms

首文本(SSE delta) - 2,108 ms 3,113 ms 2,391 ms 9,456 ms

总耗时 27,463 ms 13,896 ms 4,011 ms 15,923 ms 21,588 ms

输入 10 10 8 10 10

输出 1,861 2,055 164 1,920 1,962

缓存创建

39,254

5m: 0 / 1h: 39,254

8,058

5m: 0 / 1h: 8,058

2,069

5m: 0 / 1h: 2,069

8,860

5m: 0 / 1h: 8,860

9,171

5m: 0 / 1h: 9,171

缓存读取 0 0 8,435 0 0

输出速度 73.7 tok/s 174.3 tok/s 182.6 tok/s 141.9 tok/s 161.7 tok/s

响应

```json
{"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

```json
{"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

```json
{"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

```json
{"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

```json
{"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

步骤 6: digit_count text

时间 23:54:37 ~ 23:55:02 (21.0s) 23:52:33 ~ 23:52:43 (8.5s) 23:56:18 ~ 23:56:28 (8.1s) 23:53:28 ~ 23:54:03 (34.2s) 23:55:48 ~ 23:56:48 (58.7s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=858230388036430366446313786733241184704011649366424981173970981970814635

HTTP 头(TTFB) 1,925 ms 1,327 ms 5,900 ms 1,004 ms 20,568 ms

首文本(SSE delta) - 1,328 ms 5,901 ms 1,006 ms 20,569 ms

总耗时 19,855 ms 7,429 ms 6,503 ms 33,191 ms 56,477 ms

输入 10 193 8 193 193

输出 3,528 1,379 22 2,476 2,555

缓存创建

37,063

5m: 0 / 1h: 37,063

5m: 0 / 1h: 0

5m: 0 / 1h: 89

5m: 0 / 1h: 0

缓存读取 0 0 370 0 0

输出速度 196.8 tok/s 226.0 tok/s - 76.9 tok/s 71.2 tok/s

响应

Looking at the string `s = "858230388036430366446313786733241184704011649366424981173970981970814635"`:

**Length calculation:** 72 characters

**Digit counts:**
- Count of '0': 7 (at positions 5, 9,

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

```json
{"n":72,"c":[8,4,6]}
```

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

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

总计

总输入(含缓存) 209,779 141,519 154,931 143,054 143,265

总输出 5,883 4,033 269 5,241 5,675

总缓存创建

109,775

5m: 0 / 1h: 109,775

107,965

5m: 0 / 1h: 107,965

30,522

5m: 0 / 1h: 30,522

42,660

5m: 0 / 1h: 42,660

42,915

5m: 0 / 1h: 42,915

总缓存读取 99,944 33,311 124,361 100,151 100,107

输出速度 118.6 tok/s 184.1 tok/s 182.3 tok/s 102.4 tok/s 98.7 tok/s

总执行时间 1m 14.8s 45.3s 34.9s 1m 18.7s 2m 33.4s

总墙钟时间

15m 40.1s

含步骤间等待

13m 15.0s

含步骤间等待

17m 2.1s

含步骤间等待

14m 33.1s

含步骤间等待

17m 24.2s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1767

Anthropic: Claude Haiku 4.5

$0.1587

Anthropic: Claude Haiku 4.5

$0.0520

Anthropic: Claude Haiku 4.5

$0.0898

Anthropic: Claude Haiku 4.5

$0.0923

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count digit_count digit_count digit_count

平台 one-api (100%) one-api (100%) direct (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 - - - - -

ID 格式 msg_01V2A5Mx3ri44BAhjQD5PnFT msg_bdrk_01Dfk4UWABCYmfAxYDHdVfLk msg_BkSxvpYDj2KeK4oWKu4GBoRh msg_01VYj1NqsMorLr62JcwSc3AM msg_01HTecvaVDH74mTCeDZakf3A

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-Max	server: nginx/1.29.8 x-oneapi-request-id: 202606171554385176514768268d9d6DUNtYIDC	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-oneapi-request-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
M-Max	server: nginx x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 20260617235234415627408UqQjb7Cx	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-new-api-versionhdr:x-oneapi-request-idbody:id:msg_prefixbody:model:date_suffixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	—
M-	server: nginx	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody: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: true usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011Cc96rDxZ1vA3zsxszR4yv
cc专用	server: Nginx x-oneapi-request-id: 202606171555506929308548268d9d6uipYGG0u	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-oneapi-request-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—

协议指纹评分（机器自动）

总分 / 100 94 70 76 100 95

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 0.0 /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 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 10.0 /10 ×13 5.3 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13

原生 msg-ID 10.0 /10 ×8 0.0 /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 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 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 0.0 /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 10.0 /10 ×5

延迟基线 9.7 /10 ×5 10.0 /10 ×5 8.0 /10 ×5 8.6 /10 ×5 0.0 /10 ×5

Req-ID 透传 0.0 /10 ×4 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 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

流式投递未计入未计入未计入未计入未计入

综合结论

与基线相似度 94 v3.22.0 70 v3.22.0 76 v3.22.0 基准（参考） 95 v3.22.0

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 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 Thinking 体量 thinking_volume_match
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 11163 ms 6495 ms 4501 ms 12060 ms 24389 ms

输出速度 118.6 tok/s 184.1 tok/s 182.3 tok/s 102.4 tok/s 98.7 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 ⓘ

$0.1767

$0.1587

$0.0520

$0.0898

$0.0923

缓存读取占比 ⓘ 48% 24% 80% 70% 70%

揭晓

通道

服务商 ByteCatCode ↗

通道O-Max

服务商 SudoCode ↗

通道M-Max

服务商 AnyAIGC ↗

通道M-

官方基准

服务商 duckcoding ↗

通道cc专用

返回列表