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

2026/06/22 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

InfAI

R-Kiro

官方基准

WorldBase.ai

M-

SAIAi

O-Max

LingxiCode

M-

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 06:13:06 ~ 06:13:11 (4.6s) 06:13:14 ~ 06:13:19 (3.5s) 06:13:12 ~ 06:13:22 (5.5s) 06:13:08 ~ 06:13:13 (3.7s) 06:13:10 ~ 06:13:16 (4.4s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,441 ms 1,929 ms 3,771 ms 1,928 ms 2,735 ms

首文本(SSE delta) 2,443 ms 1,931 ms 3,772 ms 1,929 ms 2,736 ms

总耗时 2,949 ms 2,225 ms 4,178 ms 2,441 ms 3,004 ms

输入 10 10 1,696 10 10

输出 56 46 59 57 50

缓存创建

33,122

5m: 0 / 1h: 33,122

5m: 0 / 1h: 0

32,208

5m: 32,208 / 1h: 0

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

缓存读取 0 33,122 0 0 0

输出速度 110.7 tok/s 156.5 tok/s 145.3 tok/s 111.3 tok/s 186.6 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 06:15:27 ~ 06:15:37 (5.3s) 06:15:48 ~ 06:15:58 (6.8s) 06:16:24 ~ 06:16:34 (6.5s) 06:15:16 ~ 06:15:21 (4.3s) 06:16:17 ~ 06:16:27 (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) 2,318 ms 4,494 ms 3,911 ms 1,568 ms 2,775 ms

首文本(SSE delta) 2,319 ms 4,501 ms 3,912 ms 1,571 ms 2,776 ms

总耗时 4,184 ms 5,775 ms 5,305 ms 3,091 ms 4,739 ms

输入 10 10 1,706 10 10

输出 204 177 174 209 198

缓存创建

236

5m: 0 / 1h: 236

226

5m: 0 / 1h: 226

1,075

5m: 1,075 / 1h: 0

237

5m: 0 / 1h: 237

230

5m: 0 / 1h: 230

缓存读取 33,122 33,122 31,321 33,122 33,122

输出速度 109.4 tok/s 138.9 tok/s 124.9 tok/s 137.5 tok/s 100.9 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 06:18:48 ~ 06:18:53 (4.4s) 06:19:23 ~ 06:19:28 (3.5s) 06:18:35 ~ 06:18:45 (5.2s) 06:17:47 ~ 06:17:52 (3.2s) 06:18:57 ~ 06:19:02 (4.1s)

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) 2,188 ms 1,599 ms 3,588 ms 1,539 ms 2,598 ms

首文本(SSE delta) 2,189 ms 1,600 ms 3,589 ms 1,540 ms 2,599 ms

总耗时 2,810 ms 2,464 ms 4,067 ms 2,147 ms 3,171 ms

输入 10 10 1,709 10 10

输出 72 63 72 83 69

缓存创建

244

5m: 0 / 1h: 244

217

5m: 0 / 1h: 217

1,146

5m: 1,146 / 1h: 0

251

5m: 0 / 1h: 251

238

5m: 0 / 1h: 238

缓存读取 33,358 33,348 31,308 33,359 33,352

输出速度 115.9 tok/s 72.9 tok/s 150.6 tok/s 136.7 tok/s 120.6 tok/s

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 06:19:55 ~ 06:20:05 (8.0s) 06:23:14 ~ 06:23:24 (5.2s) 06:21:06 ~ 06:21:16 (7.2s) 06:20:11 ~ 06:20:21 (5.8s) 06:22:58 ~ 06:23:08 (5.9s)

提示词

查看本步 Prompt

请用中文介绍你自己是什么产品或助手，由谁开发，50 字以内。

HTTP 头(TTFB) 2,969 ms 1,618 ms 3,527 ms 1,797 ms 2,484 ms

首文本(SSE delta) 2,970 ms 1,619 ms 3,529 ms 1,798 ms 2,485 ms

总耗时 6,592 ms 4,158 ms 6,040 ms 4,716 ms 5,194 ms

输入 10 10 1,722 10 10

输出 532 376 318 463 485

缓存创建

114

5m: 0 / 1h: 114

105

5m: 0 / 1h: 105

32,707

5m: 32,707 / 1h: 0

121

5m: 0 / 1h: 121

103

5m: 0 / 1h: 103

缓存读取 33,602 33,565 0 33,610 33,590

输出速度 146.9 tok/s 148.1 tok/s 126.6 tok/s 158.7 tok/s 179.0 tok/s

响应

我是Claude，由Anthropic开发的AI助手。能帮助用户编码、调试、代码审查和解答技术问题。

我是Claude，由Anthropic公司开发的AI助手。我是Claude家族中的Haiku 4.5版本，专长于代码编写、软件工程任务、问题解答和创意写作。我可以帮助用户完成复杂的编程任务。

我是 Claude，建立在 Claude Haiku 4.5 之上。若展开一点说，我的核心定位是服务编程与工程任务的 AI 助手；关于我是谁、我是什么模型，都以 Claude 和 Claude Haiku 4.5 这组标识来理解即可。我可以帮助用户进行软件工程、代码编写、问题解答等多种任务。

我是Claude Haiku 4.5，由Anthropic开发的AI助手。作为Claude系列的轻量级版本，专为快速高效的任务设计，兼具准确性和低延迟特点。

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各种软件工程任务和编程相关工作。

步骤 5: knowledge_recall text

时间 06:21:08 ~ 06:21:33 (21.2s) 06:25:51 ~ 06:26:06 (11.8s) 06:24:56 ~ 06:25:16 (17.6s) 06:24:10 ~ 06:24:30 (17.5s) 06:24:38 ~ 06:24:53 (12.7s)

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) 1,673 ms 1,403 ms 2,516 ms 1,199 ms 2,095 ms

首文本(SSE delta) 1,675 ms 1,404 ms 2,517 ms 1,201 ms 2,096 ms

总耗时 19,679 ms 10,735 ms 16,312 ms 16,553 ms 11,878 ms

输入 10 10 450 10 10

输出 2,912 1,565 2,523 2,438 1,318

缓存创建

8,866

5m: 0 / 1h: 8,866

2,943

5m: 0 / 1h: 2,943

8,539

5m: 8,539 / 1h: 0

8,812

5m: 0 / 1h: 8,812

8,796

5m: 0 / 1h: 8,796

缓存读取 0 5,721 0 0 0

输出速度 161.7 tok/s 167.7 tok/s 182.9 tok/s 158.8 tok/s 134.7 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 06:23:44 ~ 06:23:59 (12.6s) 06:29:47 ~ 06:30:12 (21.5s) 06:27:37 ~ 06:27:58 (17.6s) 06:26:44 ~ 06:28:15 (88.1s) 06:27:52 ~ 06:28:12 (17.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=333883850023109124175562550402529598045665382412146482783497836528378428

HTTP 头(TTFB) 815 ms 936 ms 4,045 ms 1,668 ms 1,609 ms

首文本(SSE delta) 817 ms 938 ms 4,046 ms 1,670 ms 1,609 ms

总耗时 11,384 ms 20,435 ms 16,288 ms 87,194 ms 16,592 ms

输入 193 193 473 193 193

输出 2,066 4,018 1,431 16,678 2,746

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 195.5 tok/s 206.1 tok/s 116.9 tok/s 195.0 tok/s 183.3 tok/s

响应

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

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

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

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

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

总计

总输入(含缓存) 142,907 142,612 146,060 142,877 142,796

总输出 5,842 6,245 4,577 19,928 4,866

总缓存创建

42,582

5m: 0 / 1h: 42,582

3,491

5m: 0 / 1h: 3,491

75,675

5m: 75,675 / 1h: 0

42,543

5m: 0 / 1h: 42,543

42,489

5m: 0 / 1h: 42,489

总缓存读取 100,082 138,878 62,629 100,091 100,064

输出速度 166.0 tok/s 184.8 tok/s 148.5 tok/s 187.2 tok/s 160.7 tok/s

总执行时间 56.1s 52.2s 59.6s 2m 2.8s 50.8s

总墙钟时间

10m 52.8s

含步骤间等待

16m 58.0s

含步骤间等待

14m 45.1s

含步骤间等待

15m 6.1s

含步骤间等待

15m 1.2s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0927

Anthropic: Claude Haiku 4.5

$0.0497

Anthropic: Claude Haiku 4.5

$0.0960

Anthropic: Claude Haiku 4.5

$0.1631

Anthropic: Claude Haiku 4.5

$0.0877

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 (100%) direct (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 - - cloudflare - cloudflare

ID 格式 msg_01AHqbxrDUXyR6wYaPyjDwdi msg_01ABuzChyU7nsq93qFLTYKC4 msg_55e57c0e89e640b19a3460d516bf2877 msg_011y9SBJvpnUQDhfSRTxZxjm msg_01VVrpT1s3Tda8uGzeBpJFZf

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: 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_011CcHCJnbRiPmy48rBr5MR6
M-	cf-ray: a0f6711b0eb36e50-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	58415e71-8b8d-4e83-b3b6-8e36993c318f a0f6711b0eb36e50-AMS
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: 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_011CcHC5HHSzjJNGqCSVGnwy
M-	cf-ray: a0f6717288d17963-NRT server: cloudflare set-cookie: server_name_session=e44607f7628a2aebd93b10e6388b22e1; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 2026062122275449178678JUI2JLwP	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-idhdr: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	a0f6717288d17963-NRT

协议指纹评分（机器自动）

总分 / 100 98 99 69 96 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已对照基线

缓存命中比 10.0 /10 ×20 10.0 /10 ×20 3.8 /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 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 10.0 /10 ×13 7.7 /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 5.0 /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 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 10.0 /10 ×5 10.0 /10 ×5 9.6 /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 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 98 v3.24.0 基准（参考） 69 v3.24.0 96 v3.24.0 97 v3.24.0

雷达图例

共 23 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 自报身份一致 self_identity_consistency
20 信封自报一致 envelope_self_report_match
21 中文遵循 instruction_following_lang
22 档位思考量 tier_thinking_volume_match
23 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 7933 ms 7632 ms 8698 ms 19357 ms 7430 ms

输出速度 166.0 tok/s 184.8 tok/s 148.5 tok/s 187.2 tok/s 160.7 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 ⓘ

$0.0927

$0.0497

$0.0960

$0.1631

$0.0877

缓存读取占比 ⓘ 70% 97% 43% 70% 70%

揭晓

通道

服务商 InfAI ↗

通道R-Kiro

官方基准

服务商 WorldBase.ai ↗

通道M-

服务商 SAIAi ↗

通道O-Max

服务商 LingxiCode ↗

通道M-

返回列表