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-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

AiTokensFlux

O-Pro

FoxCode

R-ultra

Aiide

O-Max

官方基准

PackyCode

cc-Zeus

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 13:52:52 ~ 13:53:02 (5.4s) 13:52:58 ~ 13:53:03 (4.6s) 13:52:54 ~ 13:53:04 (6.3s) 13:53:00 ~ 13:53:10 (5.6s) 13:52:56 ~ 13:53:01 (3.9s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,037 ms 2,841 ms 4,200 ms 4,333 ms 1,758 ms

首文本(SSE delta) 3,038 ms 2,842 ms 4,201 ms 4,334 ms 1,759 ms

总耗时 3,407 ms 3,830 ms 4,709 ms 4,423 ms 1,977 ms

输入 10 33,019 10 10 10

输出 44 83 51 50 38

缓存创建

33,261

5m: 0 / 1h: 33,261

5m: 0 / 1h: 0

1,635

5m: 0 / 1h: 1,635

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 31,626 0 0

输出速度 119.2 tok/s 84.0 tok/s 100.4 tok/s - 174.3 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 13:55:54 ~ 13:56:04 (5.7s) 13:55:47 ~ 13:55:57 (5.2s) 13:55:49 ~ 13:55:59 (5.9s) 13:54:44 ~ 13:54:54 (6.5s) 13:56:47 ~ 13:56:52 (4.3s)

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) 2,666 ms 2,162 ms 3,141 ms 3,719 ms 2,076 ms

首文本(SSE delta) 2,668 ms 2,163 ms 3,142 ms 3,721 ms 2,077 ms

总耗时 4,961 ms 4,354 ms 4,648 ms 5,329 ms 3,202 ms

输入 10 33,619 10 10 10

输出 295 330 177 181 189

缓存创建

238

5m: 0 / 1h: 238

5m: 0 / 1h: 0

245

5m: 0 / 1h: 245

244

5m: 0 / 1h: 244

232

5m: 0 / 1h: 232

缓存读取 33,261 0 33,261 33,122 33,122

输出速度 128.7 tok/s 150.6 tok/s 117.5 tok/s 112.6 tok/s 168.0 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 13:57:07 ~ 13:57:12 (4.6s) 13:57:29 ~ 13:57:34 (3.8s) 13:59:56 ~ 14:00:06 (7.0s) 13:58:14 ~ 13:58:24 (5.3s) 13:58:16 ~ 13:58:21 (4.3s)

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) 2,453 ms 2,282 ms 3,948 ms 3,777 ms 1,699 ms

首文本(SSE delta) 2,454 ms 2,283 ms 3,949 ms 3,778 ms 1,700 ms

总耗时 3,475 ms 3,094 ms 4,730 ms 4,307 ms 1,876 ms

输入 10 34,317 10 10 10

输出 127 74 65 73 66

缓存创建

337

5m: 0 / 1h: 337

5m: 0 / 1h: 0

217

5m: 0 / 1h: 217

221

5m: 0 / 1h: 221

231

5m: 0 / 1h: 231

缓存读取 33,499 0 33,506 33,366 33,354

输出速度 124.4 tok/s 91.2 tok/s 83.2 tok/s 138.0 tok/s -

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 13:59:00 ~ 13:59:10 (9.4s) 13:59:54 ~ 14:00:04 (9.4s) 14:02:22 ~ 14:02:32 (6.4s) 14:01:36 ~ 14:01:46 (9.3s) 14:00:21 ~ 14:00:31 (9.7s)

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) 2,325 ms 2,048 ms 3,376 ms 3,875 ms 1,843 ms

首文本(SSE delta) 2,333 ms 2,050 ms 3,377 ms 3,877 ms 1,845 ms

总耗时 8,589 ms 8,294 ms 5,292 ms 8,284 ms 7,700 ms

输入 10 34,745 10 10 10

输出 686 1,233 314 204 802

缓存创建

167

5m: 0 / 1h: 167

5m: 0 / 1h: 0

105

5m: 0 / 1h: 105

107

5m: 0 / 1h: 107

108

5m: 0 / 1h: 108

缓存读取 33,836 0 33,723 33,587 33,585

输出速度 109.7 tok/s 197.5 tok/s 164.0 tok/s 46.3 tok/s 137.0 tok/s

响应

我叫Claude Haiku 4.5，属于Anthropic的Claude产品系列，是由Anthropic公司开发的AI助手。

我是Claude，由Anthropic公司开发的AI助手。能协助编程、写作、分析、规划等多种任务。当前版本Haiku 4.5，知识截止至2025年2月。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手模型，属于Claude产品家族。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。

我是Claude，由Anthropic开发的AI助手。帮助用户完成编程、代码审查、问题解决等软件工程任务。

步骤 5: knowledge_recall text

时间 14:00:23 ~ 14:00:53 (26.3s) 14:01:11 ~ 14:01:26 (13.6s) 14:04:03 ~ 14:04:23 (17.5s) 14:03:51 ~ 14:04:06 (13.9s) 14:03:48 ~ 14:04:28 (38.9s)

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) 1,848 ms 1,627 ms 2,628 ms 1,395 ms 2,061 ms

首文本(SSE delta) 1,850 ms 1,628 ms 2,629 ms 1,396 ms 2,062 ms

总耗时 25,520 ms 12,430 ms 16,516 ms 12,758 ms 37,620 ms

输入 10 11,321 10 10 10

输出 3,400 1,994 2,122 1,385 6,087

缓存创建

9,168

5m: 0 / 1h: 9,168

5m: 0 / 1h: 0

2,900

5m: 0 / 1h: 2,900

8,516

5m: 0 / 1h: 8,516

9,113

5m: 0 / 1h: 9,113

缓存读取 0 0 5,721 0 0

输出速度 143.6 tok/s 184.6 tok/s 152.8 tok/s 121.9 tok/s 171.2 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 14:04:18 ~ 14:05:08 (47.7s) 14:03:46 ~ 14:03:56 (9.8s) 14:06:25 ~ 14:06:45 (19.2s) 14:06:09 ~ 14:06:29 (19.1s) 14:07:09 ~ 14:07:39 (29.4s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=125038277051396933972043429741733156051855989691916466766566314953874779

HTTP 头(TTFB) 1,068 ms 1,171 ms 1,477 ms 1,189 ms 725 ms

首文本(SSE delta) 1,069 ms 1,172 ms 1,481 ms 1,191 ms 730 ms

总耗时 46,861 ms 8,978 ms 18,014 ms 18,147 ms 27,315 ms

输入 193 208 193 193 193

输出 5,883 1,681 3,172 3,169 4,635

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 128.5 tok/s 215.3 tok/s 191.9 tok/s 186.9 tok/s 174.3 tok/s

响应

```json
{"n":72,"c":[4,9,10]}
```

```json
{"n":72,"c":[4,9,10]}
```

```json
{"n":72,"c":[4,9,10]}
```

```json
{"n": 72, "c": [4, 9, 10]}
```

```json
{"n":72,"c":[4,9,10]}
```

总计

总输入(含缓存) 144,010 147,229 143,182 142,528 143,110

总输出 10,435 5,395 5,901 5,062 11,817

总缓存创建

43,171

5m: 0 / 1h: 43,171

5m: 0 / 1h: 0

5,102

5m: 0 / 1h: 5,102

42,210

5m: 0 / 1h: 42,210

42,806

5m: 0 / 1h: 42,806

总缓存读取 100,596 0 137,837 100,075 100,061

输出速度 131.4 tok/s 187.1 tok/s 168.0 tok/s 143.8 tok/s 169.5 tok/s

总执行时间 1m 39.2s 46.4s 1m 2.3s 59.7s 1m 30.5s

总墙钟时间

12m 15.9s

含步骤间等待

10m 57.2s

含步骤间等待

13m 50.2s

含步骤间等待

13m 29.0s

含步骤间等待

14m 42.5s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1164

Anthropic: Claude Haiku 4.5

$0.1742

Anthropic: Claude Haiku 4.5

$0.0499

Anthropic: Claude Haiku 4.5

$0.0883

Anthropic: Claude Haiku 4.5

$0.1228

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%) one-api (100%) one-api (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 - cloudflare

ID 格式 msg_0193nEDUTVyobu5uJf7AnG5C msg_019m0znUyYfms6LCbz4PDaG7 msg_01LxU7aZUsncsqAq7Q3N4mUY msg_01MLwqjzaiTAyNSmDcureurc msg_01EgjStCb1L1EGQCHznfqPNA

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-Pro	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	—
R-ultra	cf-ray: a12a7d81e898e393-NRT server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 20260628060347354983918268d9d6UNlnY5dp	id_format: msg_prefix 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-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	a12a7d81e898e393-NRT
O-Max	cf-ray: a12a8166afdf66ab-AMS server: cloudflare x-new-api-version: 0.12.11 x-oneapi-request-id: 202606280606265837835478268d9d6MKRShNQs	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-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	a12a8166afdf66ab-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_011CcV9xipipmByMbUogLSfc
cc-Zeus	via: 1.1 Caddy cf-ray: a12a8280ae2fb256-LAX server: cloudflare x-oneapi-request-id: 20260628140711490985199yBumXZLM	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	a12a8280ae2fb256-LAX

协议指纹评分（机器自动）

总分 / 100 93 67 97 99 96

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 0.0 /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 5.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

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

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

系统提示纯净 10.0 /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 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 10.0 /10 ×5 10.0 /10 ×5 10.0 /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 93 v3.25.0 67 v3.25.0 97 v3.25.0 基准（参考） 96 v3.25.0

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 自报身份一致 self_identity_consistency
20 信封自报一致 envelope_self_report_match
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 15469 ms 6830 ms 8985 ms 8875 ms 13282 ms

输出速度 131.4 tok/s 187.1 tok/s 168.0 tok/s 143.8 tok/s 169.5 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 ⓘ

$0.1164

$0.1742

$0.0499

$0.0883

$0.1228

缓存读取占比 ⓘ 70% 0% 96% 70% 70%

揭晓

通道

服务商 AiTokensFlux ↗

通道O-Pro

服务商 FoxCode ↗

通道R-ultra

服务商 Aiide ↗

通道O-Max

官方基准

服务商 PackyCode ↗

通道cc-Zeus

返回列表