模型对比 | RPDiag

测试说明点击展开 5 步 quick-probe-v1 详解

同一会话连续 5 步，覆盖 5 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。步与步之间随机延迟 1–4 分钟。第 3 步（cutoff）累计跨度 > 5 分钟保住 sliding cache 检测；全流程总跨度 > 6 分钟容纳新增的第 5 步。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 · 5 步骤

2026/06/15 运行日期

claude-haiku-4-5-20251001 官方基准

全部 ≥ 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

ClaudeCN

O-Max

RunAPI

O-Max

duckcoding

cc专用

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 11:02:45 ~ 11:02:55 (7.1s) 11:02:57 ~ 11:03:07 (4.9s) 11:02:51 ~ 11:02:56 (4.3s) 11:03:01 ~ 11:03:06 (3.5s) 11:02:53 ~ 11:02:58 (3.8s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,514 ms 2,775 ms 2,337 ms 2,190 ms 2,317 ms

首文本(SSE delta) 3,516 ms 2,783 ms 2,338 ms 2,192 ms 2,322 ms

总耗时 3,918 ms 3,223 ms 3,000 ms 2,542 ms 2,687 ms

输入 10 10 10 10 10

输出 52 44 54 57 48

缓存创建

33,121

5m: 0 / 1h: 33,121

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

5m: 0 / 1h: 0

33,123

5m: 0 / 1h: 33,123

缓存读取 0 0 0 33,122 0

输出速度 129.4 tok/s 100.0 tok/s 81.6 tok/s 162.9 tok/s 131.5 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 11:04:54 ~ 11:05:04 (7.6s) 11:05:45 ~ 11:05:55 (5.5s) 11:05:02 ~ 11:05:12 (6.5s) 11:06:53 ~ 11:07:03 (5.0s) 11:05:16 ~ 11:05:21 (4.7s)

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) 4,020 ms 1,568 ms 2,687 ms 1,732 ms 2,046 ms

首文本(SSE delta) 4,021 ms 1,786 ms 2,688 ms 1,733 ms 2,047 ms

总耗时 5,964 ms 3,399 ms 5,146 ms 3,697 ms 3,667 ms

输入 10 10 10 10 10

输出 230 176 262 219 194

缓存创建

246

5m: 0 / 1h: 246

238

5m: 0 / 1h: 238

248

5m: 0 / 1h: 248

251

5m: 0 / 1h: 251

242

5m: 0 / 1h: 242

缓存读取 33,121 33,122 33,122 33,122 33,123

输出速度 118.4 tok/s 109.1 tok/s 106.6 tok/s 111.5 tok/s 119.8 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 11:08:24 ~ 11:08:34 (6.2s) 11:07:47 ~ 11:07:53 (3.8s) 11:06:45 ~ 11:06:55 (5.0s) 11:08:30 ~ 11:08:35 (3.7s) 11:07:14 ~ 11:07:19 (2.8s)

提示词

查看本步 Prompt

请回报你的 knowledge cutoff，只输出 YYYY-MM。

HTTP 头(TTFB) 3,302 ms 1,700 ms 2,545 ms 2,111 ms 1,331 ms

首文本(SSE delta) 3,303 ms 1,701 ms 2,546 ms 2,113 ms 1,332 ms

总耗时 3,847 ms 2,592 ms 3,343 ms 2,612 ms 1,966 ms

输入 10 10 10 10 10

输出 62 91 92 61 63

缓存创建

260

5m: 0 / 1h: 260

218

5m: 0 / 1h: 218

287

5m: 0 / 1h: 287

259

5m: 0 / 1h: 259

217

5m: 0 / 1h: 217

缓存读取 33,367 33,360 33,370 33,373 33,365

输出速度 114.0 tok/s 102.1 tok/s 115.4 tok/s 122.2 tok/s 99.4 tok/s

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 11:11:23 ~ 11:11:33 (7.3s) 11:10:30 ~ 11:10:45 (12.1s) 11:08:04 ~ 11:08:09 (4.9s) 11:11:21 ~ 11:11:36 (9.3s) 11:09:57 ~ 11:10:12 (11.4s)

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) 3,677 ms 1,731 ms 2,545 ms 1,777 ms 1,370 ms

首文本(SSE delta) 3,680 ms 1,732 ms 2,546 ms 1,778 ms 1,371 ms

总耗时 5,968 ms 10,872 ms 4,165 ms 7,932 ms 10,340 ms

输入 10 10 10 10 10

输出 371 1,421 207 747 513

缓存创建

104

5m: 0 / 1h: 104

133

5m: 0 / 1h: 133

132

5m: 0 / 1h: 132

101

5m: 0 / 1h: 101

105

5m: 0 / 1h: 105

缓存读取 33,627 33,578 33,657 33,632 33,582

输出速度 162.2 tok/s 155.5 tok/s 127.9 tok/s 121.4 tok/s 57.2 tok/s

响应

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我能帮助进行编程、代码审查、文件编辑等各类软件工程任务。

我是由Anthropic公司开发的AI助手Claude Haiku 4.5。我协助用户进行编程、文件编辑、代码分析等软件工程任务，致力于提供安全高效的支持。

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

我是Claude Haiku 4.5，由Anthropic开发，属于Claude AI助手产品。我是一个多模态语言模型，可进行对话、编程、文本分析等工作。

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，包括编写代码、解决bug、代码审查等工作。我致力于提供安全、有用和诚实的帮助。

步骤 5: knowledge_recall text

时间 11:13:25 ~ 11:13:40 (13.6s) 11:14:08 ~ 11:14:33 (20.4s) 11:10:38 ~ 11:11:03 (20.0s) 11:13:08 ~ 11:13:33 (23.2s) 11:11:51 ~ 11:12:06 (13.1s)

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) 2,085 ms 1,180 ms 3,292 ms 1,488 ms 1,378 ms

首文本(SSE delta) 2,112 ms 1,181 ms 3,293 ms 1,489 ms 1,379 ms

总耗时 12,403 ms 18,394 ms 19,055 ms 22,210 ms 12,119 ms

输入 10 10 10 10 10

输出 1,564 2,575 2,245 3,266 1,766

缓存创建

8,720

5m: 0 / 1h: 8,720

9,750

5m: 0 / 1h: 9,750

8,614

5m: 0 / 1h: 8,614

3,377

5m: 0 / 1h: 3,377

8,818

5m: 0 / 1h: 8,818

缓存读取 0 0 0 5,721 0

输出速度 152.0 tok/s 149.6 tok/s 142.4 tok/s 157.6 tok/s 164.4 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

总计

总输入(含缓存) 142,616 143,571 142,602 143,008 142,625

总输出 2,279 4,307 2,860 4,350 2,584

总缓存创建

42,451

5m: 0 / 1h: 42,451

43,461

5m: 0 / 1h: 43,461

42,403

5m: 0 / 1h: 42,403

3,988

5m: 0 / 1h: 3,988

42,505

5m: 0 / 1h: 42,505

总缓存读取 100,115 100,060 100,149 138,970 100,070

输出速度 147.3 tok/s 147.0 tok/s 134.3 tok/s 146.5 tok/s 115.7 tok/s

总执行时间 41.8s 46.7s 40.7s 44.7s 35.8s

总墙钟时间

10m 55.4s

含步骤间等待

11m 35.8s

含步骤间等待

8m 12.1s

含步骤间等待

10m 31.9s

含步骤间等待

9m 12.8s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0745

Anthropic: Claude Haiku 4.5

$0.0859

Anthropic: Claude Haiku 4.5

$0.0774

Anthropic: Claude Haiku 4.5

$0.0407

Anthropic: Claude Haiku 4.5

$0.0761

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 knowledge_recall knowledge_recall knowledge_recall knowledge_recall knowledge_recall

平台 one-api (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

ID 格式 msg_01FcrJB9nhw2dhjRdnAD7Bsd msg_01WfaJ3svho3k4tFJNFmxG1S msg_01RnqQSUqnQUoqKp8259hYs8 msg_018XLZ2eAKPwGtP8khH9BRik msg_01FkQhHPx9F3Kfev5ad3waNn

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	cf-ray: a0be661dfe1ec129-AMS server: cloudflare x-oneapi-request-id: 20260615111327644163589UAcW8vj6	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	a0be661dfe1ec129-AMS
O-Max	server: nginx/1.24.0 (Ubuntu) x-oneapi-request-id: 20260615111410706492610tHmHm8my	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	—
cc专用	server: Nginx x-oneapi-request-id: 202606150310402630776718268d9d6SrwuYVwa	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	—
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_011Cc4KFabnsBvLmGBYpZzUn
cc-Zeus	via: 1.1 Caddy cf-ray: a0be63d0ab8abd64-LAX server: cloudflare x-oneapi-request-id: 20260615111152754315601pMcE3Tfp	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	a0be63d0ab8abd64-LAX

协议指纹评分（机器自动）

总分 / 100 95 92 95 100 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 7.4 /10 ×20 7.2 /10 ×20 7.4 /10 ×20 10.0 /10 ×20 7.4 /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 10.0 /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 9.8 /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 — — — — —

综合结论

与基线相似度 95 v3.16.0 92 v3.16.0 95 v3.16.0 基准（参考） 95 v3.16.0

雷达图例

共 21 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 中文遵循 instruction_following_lang
21 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 6420 ms 7696 ms 6942 ms 7799 ms 6156 ms

输出速度 147.3 tok/s 147.0 tok/s 134.3 tok/s 146.5 tok/s 115.7 tok/s

完成步骤 5/5 成功 5/5 成功 5/5 成功 5/5 成功 5/5 成功

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0745

$0.0859

$0.0774

$0.0407

$0.0761

缓存读取占比 ⓘ 70% 70% 70% 97% 70%

揭晓

通道

服务商 ClaudeCN ↗

通道O-Max

服务商 RunAPI ↗

通道O-Max

服务商 duckcoding ↗

通道cc专用

官方基准

服务商 PackyCode ↗

通道cc-Zeus

返回列表