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

2026/06/21 运行日期

claude-haiku-4-5-20251001 官方基准

1 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

SSSAiCode

R-Kiro

WorldBase.ai

M-

官方基准

PackyCode

cc-Zeus

ClaudeAPI

O-Bdrk

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 06:17:54 ~ 06:17:59 (4.0s) 06:17:56 ~ 06:18:06 (5.0s) 06:17:58 ~ 06:18:03 (3.9s) 06:17:50 ~ 06:17:55 (3.9s) 06:17:52 ~ 06:18:02 (6.8s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,689 ms 3,010 ms 1,994 ms 1,763 ms 4,546 ms

首文本(SSE delta) 2,707 ms 3,011 ms 1,995 ms 1,766 ms 4,548 ms

总耗时 3,179 ms 3,495 ms 2,657 ms 2,124 ms 5,039 ms

输入 0 0 10 10 10

输出 69 2 49 50 35

缓存创建

35,319

5m: 0 / 1h: 35,319

33,122

5m: 0 / 1h: 33,122

33,124

5m: 0 / 1h: 33,124

32,406

5m: 32,406 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 146.2 tok/s - 74.0 tok/s 139.7 tok/s 71.3 tok/s

响应

pong


pong

pong

pong

pong

步骤 2: identity text

时间 06:21:09 ~ 06:21:14 (4.9s) 06:21:52 ~ 06:22:02 (5.8s) 06:21:42 ~ 06:21:52 (6.4s) 06:19:15 ~ 06:19:20 (4.2s) 06:20:27 ~ 06:20:37 (9.2s)

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) 2,306 ms 2,922 ms 2,094 ms 1,555 ms 3,023 ms

首文本(SSE delta) 2,327 ms 2,923 ms 2,097 ms 1,556 ms 3,107 ms

总耗时 4,018 ms 4,485 ms 5,334 ms 3,142 ms 7,912 ms

输入 0 211 10 10 10

输出 262 26 234 190 216

缓存创建

5m: 0 / 1h: 48

240

5m: 0 / 1h: 240

241

5m: 0 / 1h: 241

189

5m: 189 / 1h: 0

缓存读取 0 35,296 33,122 33,124 32,406

输出速度 154.9 tok/s - 72.3 tok/s 119.8 tok/s 45.0 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 06:24:05 ~ 06:24:10 (3.9s) 06:23:59 ~ 06:24:04 (4.7s) 06:25:23 ~ 06:25:28 (3.7s) 06:20:53 ~ 06:20:58 (3.3s) 06:24:30 ~ 06:24:40 (6.1s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 2,433 ms 2,851 ms 1,780 ms 1,573 ms 3,386 ms

首文本(SSE delta) 2,441 ms 2,852 ms 1,781 ms 1,574 ms 3,462 ms

总耗时 2,991 ms 3,535 ms 2,623 ms 2,114 ms 4,763 ms

输入 0 41 10 10 10

输出 65 5 69 64 60

缓存创建

470

5m: 0 / 1h: 470

276

5m: 0 / 1h: 276

213

5m: 0 / 1h: 213

5m: 42 / 1h: 0

缓存读取 0 35,320 33,362 33,365 32,595

输出速度 118.2 tok/s - 81.9 tok/s 118.5 tok/s 46.1 tok/s

响应

2025-02


2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 06:25:38 ~ 06:25:48 (5.6s) 06:27:02 ~ 06:27:12 (7.1s) 06:28:47 ~ 06:28:57 (8.1s) 06:23:51 ~ 06:24:01 (5.3s) 06:26:04 ~ 06:26:14 (8.2s)

提示词

查看本步 Prompt

用中文说说你的身份：什么产品？由谁打造？50 字以内。

HTTP 头(TTFB) 2,400 ms 3,020 ms 2,522 ms 2,408 ms 2,780 ms

首文本(SSE delta) 2,403 ms 3,026 ms 2,524 ms 2,409 ms 2,781 ms

总耗时 4,813 ms 6,069 ms 6,934 ms 4,232 ms 6,860 ms

输入 0 39 10 10 10

输出 279 27 370 273 327

缓存创建

125

5m: 0 / 1h: 125

111

5m: 0 / 1h: 111

104

5m: 0 / 1h: 104

5m: 35 / 1h: 0

缓存读取 0 35,778 33,638 33,578 32,637

输出速度 115.8 tok/s - 83.9 tok/s 149.8 tok/s 80.2 tok/s

响应

我是Claude Haiku 4.5，由Anthropic公司打造的AI助手。我专长于代码编写、软件工程任务和技术问题解答。


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

我是Claude Haiku 4.5，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、问题解答等多种工作。

我叫Claude Haiku 4.5，属于Claude产品家族，由Anthropic公司开发。

我是Claude Haiku，由Anthropic公司开发的AI助手。我可以帮助您进行编程、分析和各类信息任务。

步骤 5: knowledge_recall text

时间 06:27:49 ~ 06:28:00 (8.7s) 06:29:35 ~ 06:29:51 (14.0s) 06:30:04 ~ 06:30:24 (18.4s) 06:26:33 ~ 06:26:58 (21.8s) 06:29:18 ~ 06:29:59 (35.2s)

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) 1,873 ms 1,849 ms 1,516 ms 1,889 ms 2,581 ms

首文本(SSE delta) 1,880 ms 1,850 ms 1,517 ms 1,893 ms 2,656 ms

总耗时 7,889 ms 12,698 ms 17,380 ms 20,716 ms 33,949 ms

输入 0 1,003 10 10 10

输出 1,254 176 1,489 2,762 2,850

缓存创建

9,372

5m: 0 / 1h: 9,372

8,737

5m: 0 / 1h: 8,737

8,573

5m: 0 / 1h: 8,573

8,045

5m: 8,045 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 208.7 tok/s 16.2 tok/s 93.9 tok/s 146.7 tok/s 91.1 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":"Spa

```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

{"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

```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 06:31:29 ~ 06:31:39 (8.6s) 06:33:15 ~ 06:33:30 (11.9s) 06:32:25 ~ 06:32:45 (17.8s) 06:30:18 ~ 06:30:53 (31.0s) 06:31:25 ~ 06:32:35 (66.4s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=259433196961352631864027441906201012582203127717390854296190723773181597

HTTP 头(TTFB) 1,467 ms 1,512 ms 1,387 ms 955 ms 1,503 ms

首文本(SSE delta) 1,478 ms 1,513 ms 1,388 ms 956 ms 1,505 ms

总耗时 7,702 ms 10,823 ms 16,703 ms 29,883 ms 65,084 ms

输入 0 198 193 193 193

输出 1,052 18 2,731 5,902 11,767

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 169.0 tok/s - 178.3 tok/s 204.0 tok/s 185.1 tok/s

响应

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


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

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

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

```json
{
  "n": 74,
  "c": [7, 8, 9]
}
```

总计

总输入(含缓存) 0 153,220 142,851 142,565 138,598

总输出 2,981 254 4,942 9,241 15,255

总缓存创建

45,334

5m: 0 / 1h: 45,334

42,486

5m: 0 / 1h: 42,486

42,255

5m: 0 / 1h: 42,255

40,717

5m: 40,717 / 1h: 0

总缓存读取 0 106,394 100,122 100,067 97,638

输出速度 171.8 tok/s 16.2 tok/s 122.5 tok/s 177.5 tok/s 144.5 tok/s

总执行时间 35.7s 48.5s 58.3s 1m 9.4s 2m 11.9s

总墙钟时间

13m 45.0s

含步骤间等待

15m 34.2s

含步骤间等待

14m 47.0s

含步骤间等待

13m 3.0s

含步骤间等待

14m 43.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

>=$0.0149

Anthropic: Claude Haiku 4.5 (成本不完整)

$0.0701

Anthropic: Claude Haiku 4.5

$0.0881

Anthropic: Claude Haiku 4.5

$0.1093

Anthropic: Claude Haiku 4.5

$0.1181

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 direct (75%) direct (100%) direct (100%) one-api (100%) one-api (100%)

上游 Anthropic 直连 (75%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%)

CDN - cloudflare - cloudflare -

ID 格式 msg_d48611bfb8ed4f998f90acfbe5717df6 msg_01cPJ8jbKMCF13cXBHJP4ysg msg_01P8iqhMkvKmQkYjDRuj6mGm msg_01AX59AkGYo5DRkMwvC6H4zy msg_01HdH5kf1zDmuHUuE1LR1AZy

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
R-Kiro	server: Photon-Edge x-amzn-requestid: f6d74fd4-7325-4bb1-9164-7ce5d131cf24	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.cache_fieldsupstream:x-amzn-requestidupstream:anthropic-direct-from-body	1781994690283-ithy9tphbm6ap f6d74fd4-7325-4bb1-9164-7ce5d131cf24
M-	cf-ray: a0ee3bf5fc8d0e36-AMS server: cloudflare	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:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	ebf01ff4-a1e2-4419-b289-253f40feee70 a0ee3bf5fc8d0e36-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: 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_011CcFJhTuNHF3MGMQwk1PSV
cc-Zeus	via: 1.1 Caddy cf-ray: a0ee37a26c00b860-LAX server: cloudflare x-oneapi-request-id: 20260621063019289286192JYvsWPlV	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: true usage_has_cache_fields: true	hdr:x-oneapi-request-idhdr:server:cloudflarehdr:cf-rayhdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a0ee37a26c00b860-LAX
O-Bdrk	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606202231265291190308268d9d6QTI6Mgbn	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-new-api-versionhdr: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 51 85 100 98 85

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 10.0 /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 0.0 /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 10.0 /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

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 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 10.0 /10 ×5 10.0 /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 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

系统提示纯净未计入 0.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 5.0 /10 ×8

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 51 v3.24.0 85 v3.24.0 基准（参考） 98 v3.24.0 85 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 自由身份 identity_free_clean
6 停止原因 stop_reason_present
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 Thinking 体量 thinking_volume_match
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 自报身份一致 self_identity_consistency
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 5099 ms 6851 ms 8605 ms 10369 ms 20601 ms

输出速度 171.8 tok/s 16.2 tok/s 122.5 tok/s 177.5 tok/s 144.5 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 ⓘ

>=$0.0149 (成本不完整)

$0.0701

$0.0881

$0.1093

$0.1181

缓存读取占比 ⓘ - 69% 70% 70% 70%

揭晓

通道

服务商 SSSAiCode ↗

通道R-Kiro

服务商 WorldBase.ai ↗

通道M-

官方基准

服务商 PackyCode ↗

通道cc-Zeus

服务商 ClaudeAPI ↗

通道O-Bdrk

返回列表