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

1 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

官方基准

FastCode

ClaudeAPI

O-Bdrk

InfAI

R-Kiro

ClaudeZZ

R-kiro

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 14:21:38 ~ 14:21:43 (3.8s) 14:21:30 ~ 14:21:50 (17.6s) 14:21:32 ~ 14:21:42 (5.6s) 14:21:34 ~ 14:21:44 (6.5s) 14:21:36 ~ 14:21:56 (17.1s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,195 ms 14,319 ms 2,421 ms 4,023 ms 15,622 ms

首文本(SSE delta) 2,198 ms 14,323 ms 2,423 ms 4,024 ms 15,624 ms

总耗时 2,561 ms 15,303 ms 3,112 ms 4,369 ms 15,998 ms

输入 10 10 10 10 10

输出 54 68 39 46 53

缓存创建

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

22,404

5m: 0 / 1h: 22,404

缓存读取 0 0 0 0 0

输出速度 148.8 tok/s 69.4 tok/s 56.6 tok/s 133.3 tok/s 141.7 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 14:23:06 ~ 14:23:11 (4.9s) 14:24:35 ~ 14:24:45 (9.0s) 14:24:16 ~ 14:24:26 (6.5s) 14:24:05 ~ 14:24:15 (7.9s) 14:25:35 ~ 14:25:50 (11.0s)

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。请按下列 3 行格式精确填空作答（label 不变，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) 1,829 ms 3,118 ms 2,856 ms 3,345 ms 3,723 ms

首文本(SSE delta) 1,830 ms 3,626 ms 2,857 ms 3,346 ms 3,725 ms

总耗时 3,556 ms 7,695 ms 4,489 ms 6,484 ms 9,896 ms

输入 10 10 10 10 10

输出 243 221 206 246 592

缓存创建

238

5m: 0 / 1h: 238

182

5m: 0 / 1h: 182

182

5m: 0 / 1h: 182

230

5m: 0 / 1h: 230

244

5m: 0 / 1h: 244

缓存读取 33,122 33,122 33,122 33,122 22,404

输出速度 140.8 tok/s 54.3 tok/s 126.2 tok/s 78.4 tok/s 95.9 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 14:25:00 ~ 14:25:05 (4.0s) 14:28:43 ~ 14:28:53 (6.3s) 14:26:41 ~ 14:26:46 (4.6s) 14:25:24 ~ 14:25:34 (7.5s) 14:29:03 ~ 14:29:13 (5.3s)

提示词

查看本步 Prompt

训练数据截止月份是？只回 YYYY-MM。

HTTP 头(TTFB) 2,459 ms 3,304 ms 1,707 ms 2,939 ms 3,883 ms

首文本(SSE delta) 2,460 ms 4,315 ms 1,708 ms 2,940 ms 3,885 ms

总耗时 3,027 ms 5,075 ms 2,391 ms 4,091 ms 4,459 ms

输入 10 10 10 10 10

输出 65 81 66 70 63

缓存创建

272

5m: 0 / 1h: 272

5m: 0 / 1h: 44

5m: 0 / 1h: 46

275

5m: 0 / 1h: 275

649

5m: 0 / 1h: 649

缓存读取 33,360 33,304 33,304 33,352 22,648

输出速度 114.6 tok/s 106.6 tok/s 96.6 tok/s 60.8 tok/s 109.8 tok/s

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 14:26:53 ~ 14:27:03 (8.0s) 14:31:34 ~ 14:31:44 (8.2s) 14:29:16 ~ 14:29:31 (10.1s) 14:27:27 ~ 14:27:37 (5.5s) 14:33:13 ~ 14:33:23 (8.8s)

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) 1,848 ms 3,949 ms 1,655 ms 2,929 ms 4,190 ms

首文本(SSE delta) 1,850 ms 3,950 ms 1,656 ms 2,930 ms 4,192 ms

总耗时 7,015 ms 6,927 ms 7,702 ms 4,259 ms 8,042 ms

输入 10 10 10 10 10

输出 717 185 787 161 628

缓存创建

103

5m: 0 / 1h: 103

5m: 0 / 1h: 41

5m: 0 / 1h: 39

112

5m: 0 / 1h: 112

115

5m: 0 / 1h: 115

缓存读取 33,632 33,348 33,350 33,627 23,297

输出速度 138.8 tok/s 62.1 tok/s 130.2 tok/s 121.1 tok/s 163.1 tok/s

响应

我是 Claude Haiku 4.5，由 Anthropic 公司打造的 AI 助手。是 Claude 系列中最快的模型，专长于快速高效地完成各类任务。

我是 Claude Haiku 4.5，由 Anthropic 公司开发的 AI 助手。当前在 Claude Code 环境中运行，帮助用户完成软件开发任务。

我是Claude，由Anthropic开发的AI助手。当前运行的是Claude Haiku 4.5模型。

我是Claude Code，由Anthropic开发的AI编程助手。我可以帮助软件工程任务，包括解bug、新增功能、代码重构和解释代码等。我能读写文件、运行命令、执行测试。

我是Claude，由Anthropic开发的AI助手，专长于软件工程和编程任务。

步骤 5: knowledge_recall text

时间 14:29:56 ~ 14:30:17 (19.8s) 14:34:58 ~ 14:35:18 (19.1s) 14:32:33 ~ 14:33:08 (31.6s) 14:30:05 ~ 14:30:50 (44.0s) 14:36:29 ~ 14:36:49 (17.6s)

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) 1,562 ms 3,153 ms 1,815 ms 13,315 ms 4,266 ms

首文本(SSE delta) 1,564 ms 3,154 ms 1,816 ms 13,316 ms 4,267 ms

总耗时 18,860 ms 16,319 ms 29,991 ms 41,781 ms 16,760 ms

输入 10 10 10 10 10

输出 2,958 1,057 4,598 4,630 2,111

缓存创建

9,070

5m: 0 / 1h: 9,070

8,060

5m: 0 / 1h: 8,060

8,036

5m: 0 / 1h: 8,036

8,518

5m: 0 / 1h: 8,518

9,511

5m: 0 / 1h: 9,511

缓存读取 0 0 0 0 0

输出速度 171.0 tok/s 80.3 tok/s 163.2 tok/s 162.7 tok/s 169.0 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) 142,969 141,273 141,251 142,408 101,322

总输出 4,037 1,612 5,696 5,153 3,447

总缓存创建

42,805

5m: 0 / 1h: 42,805

41,449

5m: 0 / 1h: 41,449

41,425

5m: 0 / 1h: 41,425

42,257

5m: 0 / 1h: 42,257

32,923

5m: 0 / 1h: 32,923

总缓存读取 100,114 99,774 99,776 100,101 68,349

输出速度 160.7 tok/s 73.4 tok/s 153.0 tok/s 149.7 tok/s 146.9 tok/s

总执行时间 40.6s 1m 0.1s 58.4s 1m 11.4s 59.8s

总墙钟时间

8m 38.8s

含步骤间等待

13m 48.8s

含步骤间等待

11m 36.9s

含步骤间等待

9m 16.9s

含步骤间等待

15m 13.8s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0838

Anthropic: Claude Haiku 4.5

$0.0699

Anthropic: Claude Haiku 4.5

$0.0903

Anthropic: Claude Haiku 4.5

$0.0886

Anthropic: Claude Haiku 4.5

$0.0653

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 direct (100%) direct (100%) one-api (100%) direct (100%) direct (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_01C1dTZWhRyf3fY4azA5oEVX msg_vrtx_01Xf3hU2UW9QuP8LfiZBFaUJ msg_01AjWTQ2jTx34wssfzAqt11m msg_01TVfrty6EqFjsmmtid3Z46M msg_019j9neohMynzXhtSjB9bqUj

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
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_011Cc4aG2bUinYw7M36Q2zuR
cc	server: nginx	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	body:id:msg_prefixbody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
O-Bdrk	x-new-api-version: v2026.05.18 x-oneapi-request-id: 202606150632355342517668268d9d6bhOpHES0	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	—
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	—
R-kiro	via: 1.1 google cf-ray: 4c20bf94cbd4171bc302c4314c636e1e-IAD server: nginx	id_format: msg_prefix via_chain: ["1.1 google"] extra_fields: ["diagnostics","stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_4c20bf94cbd4171bc302c4314c636e1e 4c20bf94cbd4171bc302c4314c636e1e-IAD

协议指纹评分（机器自动）

总分 / 100 96 80 95 95 88

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 7.3 /10 ×20 7.6 /10 ×20 7.6 /10 ×20 7.4 /10 ×20 6.5 /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 0.0 /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 0.0 /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 10.0 /10 ×5 9.2 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 7.0 /10 ×5

Req-ID 透传 10.0 /10 ×4 0.0 /10 ×4 0.0 /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

流式投递 — — — — —

综合结论

与基线相似度基准（参考） 80 v3.16.0 95 v3.16.0 95 v3.16.0 88 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 7004 ms 10264 ms 9537 ms 12197 ms 11031 ms

输出速度 160.7 tok/s 73.4 tok/s 153.0 tok/s 149.7 tok/s 146.9 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 ⓘ

$0.0838

$0.0699

$0.0903

$0.0886

$0.0653

缓存读取占比 ⓘ 70% 71% 71% 70% 67%

揭晓

通道

官方基准

服务商 FastCode ↗

通道cc

服务商 ClaudeAPI ↗

通道O-Bdrk

服务商 InfAI ↗

通道R-Kiro

服务商 ClaudeZZ ↗

通道R-kiro

返回列表