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

2026/06/30 运行日期

claude-haiku-4-5-20251001 官方基准

1 任务失败执行不完整

已揭晓揭晓进度

步骤 / 指标

ClaudeZZ

R-kiro

SSSAiCode

R-Kiro

FastCode

SudoCode

M-Max

官方基准

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 13:11:15 ~ 13:11:25 (5.8s) 13:11:11 ~ 13:11:26 (13.0s) 13:11:09 ~ 13:11:19 (5.7s) 13:11:13 ~ 13:11:18 (4.0s) 13:11:17 ~ 13:11:22 (4.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,878 ms 1,661 ms 3,149 ms 2,180 ms 2,457 ms

首文本(SSE delta) 3,880 ms 1,707 ms 3,152 ms 2,182 ms 2,459 ms

总耗时 4,466 ms 3,043 ms 3,276 ms 2,756 ms 2,930 ms

输入 10 0 10 10 10

输出 55 178 50 62 50

缓存创建

22,404

5m: 0 / 1h: 22,404

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 0 0 0

输出速度 93.9 tok/s 133.2 tok/s - 108.0 tok/s 106.2 tok/s

响应

pong

I'm ready. What would you like me to work on?

pong

pong

pong

步骤 2: identity text

时间 13:14:20 ~ 13:14:30 (6.9s) 13:13:17 ~ 13:13:27 (5.5s) 13:13:38 ~ 13:13:48 (8.8s) 13:13:57 ~ 13:14:02 (4.7s) 13:12:51 ~ 13:13:01 (5.3s)

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) 3,393 ms 2,733 ms 4,179 ms 1,678 ms 1,703 ms

首文本(SSE delta) 3,395 ms 2,734 ms 4,180 ms 1,679 ms 1,705 ms

总耗时 5,845 ms 4,819 ms 7,214 ms 3,494 ms 4,355 ms

输入 10 0 10 10 10

输出 214 303 296 200 259

缓存创建

253

5m: 0 / 1h: 253

241

5m: 0 / 1h: 241

253

5m: 0 / 1h: 253

241

5m: 0 / 1h: 241

缓存读取 22,404 0 33,122 33,122 33,122

输出速度 87.3 tok/s 145.3 tok/s 97.6 tok/s 110.2 tok/s 97.7 tok/s

响应

vendor: anthropic
brand: claude
model: claude-haiku-4-5

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 13:18:01 ~ 13:18:12 (5.6s) 13:14:48 ~ 13:15:08 (15.7s) 13:16:15 ~ 13:16:30 (14.9s) 13:16:10 ~ 13:16:16 (4.0s) 13:15:39 ~ 13:15:45 (3.9s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 3,711 ms 1,947 ms 12,963 ms 2,644 ms 2,272 ms

首文本(SSE delta) 3,713 ms 1,952 ms 12,964 ms 2,688 ms 2,274 ms

总耗时 4,627 ms 4,603 ms 13,530 ms 3,040 ms 2,898 ms

输入 10 0 10 10 10

输出 63 414 78 66 62

缓存创建

261

5m: 0 / 1h: 261

336

5m: 0 / 1h: 336

240

5m: 0 / 1h: 240

282

5m: 0 / 1h: 282

缓存读取 22,657 0 33,363 33,375 33,363

输出速度 68.9 tok/s - 137.8 tok/s 187.5 tok/s 99.4 tok/s

响应

2025-02

失败

评测失败

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 13:20:07 ~ 13:20:17 (8.7s) - 13:18:58 ~ 13:19:08 (5.2s) 13:19:48 ~ 13:19:58 (5.4s) 13:16:47 ~ 13:16:57 (5.0s)

提示词

查看本步 Prompt

中文做个自我介绍：你是什么产品或助手？谁开发的？50 字内。

HTTP 头(TTFB) 4,175 ms - 2,719 ms 2,995 ms 1,431 ms

首文本(SSE delta) 4,177 ms - 2,720 ms 3,052 ms 1,433 ms

总耗时 7,792 ms - 4,087 ms 4,325 ms 3,950 ms

输入 10 - 10 10 10

输出 434 - 280 176 267

缓存创建

115

5m: 0 / 1h: 115

118

5m: 0 / 1h: 118

104

5m: 0 / 1h: 104

102

5m: 0 / 1h: 102

缓存读取 22,918 - 33,699 33,615 33,645

输出速度 120.1 tok/s - 204.8 tok/s 138.3 tok/s 106.1 tok/s

响应

我是Claude Code，由Anthropic开发的AI编程助手。我可以帮助用户进行软件工程任务，包括解决bug、添加功能、重构代码等。

未执行

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公司开发的AI助手。我是Claude家族中较轻量的模型，擅长快速处理任务。

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

步骤 5: knowledge_recall text

时间 13:22:37 ~ 13:23:07 (25.4s) - 13:20:09 ~ 13:20:39 (25.4s) 13:21:20 ~ 13:21:50 (28.8s) 13:20:05 ~ 13:20:25 (17.4s)

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) 3,520 ms - 2,091 ms 1,651 ms 1,850 ms

首文本(SSE delta) 3,539 ms - 2,092 ms 1,653 ms 1,853 ms

总耗时 24,611 ms - 22,712 ms 27,788 ms 16,463 ms

输入 10 - 10 10 10

输出 2,476 - 3,471 3,843 1,769

缓存创建

8,962

5m: 0 / 1h: 8,962

8,715

5m: 0 / 1h: 8,715

8,513

5m: 0 / 1h: 8,513

8,632

5m: 0 / 1h: 8,632

缓存读取 0 - 0 0 0

输出速度 117.5 tok/s - 168.3 tok/s 147.0 tok/s 121.1 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

未执行

```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 13:25:34 ~ 13:25:59 (21.6s) - 13:21:50 ~ 13:22:05 (14.8s) 13:23:52 ~ 13:24:12 (18.0s) 13:24:07 ~ 13:24:37 (28.2s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=359829208774977435507957516852097699794512639133878665498263202854583435

HTTP 头(TTFB) 2,169 ms - 1,564 ms 2,842 ms 749 ms

首文本(SSE delta) 2,170 ms - 1,565 ms 2,843 ms 751 ms

总耗时 20,868 ms - 13,604 ms 16,990 ms 27,189 ms

输入 10 - 193 193 193

输出 2,457 - 2,142 2,112 5,824

缓存创建

5,980

5m: 0 / 1h: 5,980

5m: 0 / 1h: 0

缓存读取 0 - 0 0 0

输出速度 131.4 tok/s - 177.9 tok/s 149.3 tok/s 220.3 tok/s

响应

I'll count the digits in the string exactly.

```
s = "359829208774977435507957516852097699794512639133878665498263202854583435"
```

Counting:
- **Length (n)**: 72 characters
- **Count of 0s**: 4 (po

未执行

```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)

106,014

仅前 3 步

142,959

142,587

142,752

总输出

5,699

895

仅前 3 步

6,317

6,459

8,231

总缓存创建

37,975

5m: 0 / 1h: 37,975

仅前 3 步

42,532

5m: 0 / 1h: 42,532

42,232

5m: 0 / 1h: 42,232

42,379

5m: 0 / 1h: 42,379

总缓存读取

67,979

仅前 3 步

100,184

100,112

100,130

输出速度

120.4 tok/s

140.6 tok/s

仅前 3 步

166.6 tok/s

145.8 tok/s

174.0 tok/s

总执行时间

1m 13.9s

34.3s

仅前 3 步

1m 14.8s

1m 4.9s

1m 4.1s

总墙钟时间

14m 44.2s

含步骤间等待

3m 56.5s

含步骤间等待

仅前 3 步

10m 56.0s

含步骤间等待

12m 59.1s

含步骤间等待

13m 20.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0828

Anthropic: Claude Haiku 4.5

>=$0.0045

Anthropic: Claude Haiku 4.5 (成本不完整 · 部分)

$0.0950

Anthropic: Claude Haiku 4.5

$0.0953

Anthropic: Claude Haiku 4.5

$0.1044

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count cutoff digit_count digit_count digit_count

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

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

CDN - - - - -

ID 格式 msg_019VfncpB99y23gWbovT1N8w msg_8d39a7cfcbc040fda67eafc1249ccd68 msg_01DEzweeb6cGScBfWdwg3i8K msg_01MQyZyHPZt2ZDGERsfARbRx msg_01NfTmsSPvvasZ688PjxEDDu

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	—
R-Kiro	server: Photon-Edge x-amzn-requestid: b90e30e3-9782-49fc-8113-3d9c9502975d	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	1782796499327-yrndhu7rhcyoe b90e30e3-9782-49fc-8113-3d9c9502975d
cc	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	—
M-Max	server: nginx x-new-api-version: 0b8be5cf x-oneapi-request-id: 2026063013235372773824519eoFYWp	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	—
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_011CcYtNcexwhUSpbkc4ECdg

协议指纹评分（机器自动）

总分 / 100 93 97 98 100

基线状态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 10.0 /10 ×20 10.0 /10 ×20

缓存 TTL 一致 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

缓存连续性 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

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

系统提示纯净 0.0 /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

知识截止 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

service_tier 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

延迟基线 8.8 /10 ×5 — 7.5 /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

stop_reason 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

流式投递未计入 — 未计入未计入未计入

综合结论

与基线相似度 93 v3.25.0 执行失败，不评分 97 v3.25.0 98 v3.25.0 基准（参考）

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 11368 ms 3931 ms 10737 ms 9732 ms 9631 ms

输出速度 120.4 tok/s 140.6 tok/s 166.6 tok/s 145.8 tok/s 174.0 tok/s

完成步骤 6/6 成功 2/6 成功 · 1 失败 · 3 跳过 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 ⓘ

$0.0828

>=$0.0045 (成本不完整 · 部分)

$0.0950

$0.0953

$0.1044

缓存读取占比 ⓘ 64% - (部分) 70% 70% 70%

揭晓

通道

服务商 ClaudeZZ ↗

通道R-kiro

服务商 SSSAiCode ↗

通道R-Kiro

服务商 FastCode ↗

通道cc

服务商 SudoCode ↗

通道M-Max

官方基准

返回列表