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

2026/06/23 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

LingxiCode

M-

InfAI

R-Kiro

ClaudeZZ

R-kiro

AnyAIGC

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 09:00:24 ~ 09:00:29 (3.4s) 09:00:18 ~ 09:00:28 (5.2s) 09:00:16 ~ 09:00:26 (4.9s) 09:00:22 ~ 09:00:27 (4.9s) 09:00:20 ~ 09:00:25 (4.5s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 1,926 ms 3,200 ms 3,247 ms 3,103 ms 2,854 ms

首文本(SSE delta) 1,928 ms 3,235 ms 3,250 ms 3,105 ms 2,855 ms

总耗时 2,202 ms 3,559 ms 3,252 ms 3,686 ms 2,856 ms

输入 10 10 10 10 8,600

输出 43 47 49 47 2

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

22,403

5m: 0 / 1h: 22,403

缓存读取 0 0 0 0 0

输出速度 156.9 tok/s 145.1 tok/s burst 80.9 tok/s -

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 09:01:32 ~ 09:01:42 (7.4s) 09:02:08 ~ 09:02:18 (6.6s) 09:03:30 ~ 09:03:40 (6.0s) 09:02:33 ~ 09:02:43 (5.8s) 09:02:18 ~ 09:02:28 (6.7s)

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。严格按以下 3 行格式输出，全部小写英文，不要任何其他文字、解释、标点：
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) 3,973 ms 3,306 ms 2,850 ms 3,065 ms 3,342 ms

首文本(SSE delta) 3,974 ms 3,308 ms 2,851 ms 3,091 ms 3,343 ms

总耗时 6,337 ms 5,483 ms 4,808 ms 4,946 ms 4,405 ms

输入 10 10 10 10 8,801

输出 200 241 277 194 25

缓存创建

236

5m: 0 / 1h: 236

240

5m: 0 / 1h: 240

242

5m: 0 / 1h: 242

247

5m: 0 / 1h: 247

缓存读取 33,122 33,122 33,122 22,403 0

输出速度 84.6 tok/s 110.8 tok/s 141.5 tok/s 104.6 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 09:02:59 ~ 09:03:05 (3.6s) 09:06:13 ~ 09:06:23 (5.1s) 09:04:42 ~ 09:04:47 (4.6s) 09:04:59 ~ 09:05:04 (4.5s) 09:06:10 ~ 09:06:21 (6.3s)

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) 2,103 ms 3,207 ms 3,373 ms 2,852 ms 4,641 ms

首文本(SSE delta) 2,105 ms 3,208 ms 3,374 ms 2,853 ms 4,642 ms

总耗时 2,586 ms 3,813 ms 3,375 ms 3,682 ms 4,803 ms

输入 10 10 10 10 8,841

输出 70 79 65 67 5

缓存创建

240

5m: 0 / 1h: 240

283

5m: 0 / 1h: 283

300

5m: 0 / 1h: 300

264

5m: 0 / 1h: 264

缓存读取 33,358 33,362 33,364 22,650 0

输出速度 145.5 tok/s 130.6 tok/s burst 80.8 tok/s -

响应

2025-02

2025-02

2025-02

2025-02

2025-05

步骤 4: identity_free text

时间 09:04:06 ~ 09:04:16 (8.5s) 09:08:10 ~ 09:08:25 (10.1s) 09:08:39 ~ 09:08:54 (13.4s) 09:08:33 ~ 09:08:53 (15.7s) 09:09:46 ~ 09:09:56 (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) 2,421 ms 2,582 ms 2,888 ms 3,425 ms 3,293 ms

首文本(SSE delta) 2,422 ms 2,585 ms 2,889 ms 3,427 ms 3,294 ms

总耗时 7,426 ms 9,313 ms 11,649 ms 14,787 ms 3,657 ms

输入 10 10 10 10 8,873

输出 622 1,095 1,449 1,874 23

缓存创建

112

5m: 0 / 1h: 112

121

5m: 0 / 1h: 121

107

5m: 0 / 1h: 107

117

5m: 0 / 1h: 117

缓存读取 33,598 33,645 33,664 22,914 0

输出速度 124.3 tok/s 162.8 tok/s 165.4 tok/s 165.0 tok/s -

响应

我是Claude Haiku 4.5，由Anthropic开发的AI助手。可帮助代码编辑、文本分析、问题解答等多种任务。

我是Claude Code，Anthropic开发的AI编程助手，可帮助完成代码编写、调试、测试和重构等软件开发任务。

我是Claude，由Anthropic公司开发的AI助手。可帮助编程、代码分析、文本处理等任务。当前运行Haiku 4.5模型。

我叫Claude Haiku 4.5，由Anthropic开发，属Claude产品，是AI编程助手。

我是Claude，由Anthropic开发的人工智能助手，用于帮助解答和处理软件工程相关问题。

步骤 5: knowledge_recall text

时间 09:05:51 ~ 09:06:06 (14.7s) 09:11:01 ~ 09:11:21 (19.7s) 09:10:51 ~ 09:11:06 (13.3s) 09:11:19 ~ 09:11:34 (14.2s) 09:12:44 ~ 09:12:54 (5.4s)

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) 2,494 ms 3,012 ms 2,396 ms 2,593 ms 3,582 ms

首文本(SSE delta) 2,496 ms 3,014 ms 2,398 ms 2,595 ms 3,583 ms

总耗时 13,688 ms 18,531 ms 12,093 ms 13,326 ms 3,835 ms

输入 10 10 10 10 9,503

输出 1,875 2,939 1,564 1,887 154

缓存创建

8,950

5m: 0 / 1h: 8,950

9,479

5m: 0 / 1h: 9,479

9,838

5m: 0 / 1h: 9,838

10,379

5m: 0 / 1h: 10,379

缓存读取 0 0 0 0 0

输出速度 167.5 tok/s 189.4 tok/s 161.3 tok/s 175.8 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_20

```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

{"facts":{"super_bowl_lviii":"UNKNOWN","oscars2024_best_picture":"UNKNOWN","ucl_2024":"UNKNOWN","nba_2024":"UNKNOWN","copa_america_2024":"UNKNOWN","euro2024":"UNKNOWN","nobel_peace_2024":"UNKNOWN","wo

步骤 6: digit_count text

时间 09:08:55 ~ 09:09:15 (15.1s) 09:14:17 ~ 09:14:32 (14.2s) 09:12:34 ~ 09:13:09 (30.2s) 09:14:56 ~ 09:15:16 (16.6s) 09:15:16 ~ 09:15:26 (6.2s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=607467364493081224916910171690428524148292534530601440814950871454645956

HTTP 头(TTFB) 1,379 ms 1,581 ms 2,922 ms 2,943 ms 3,435 ms

首文本(SSE delta) 1,381 ms 1,582 ms 3,223 ms 2,944 ms 3,436 ms

总耗时 14,059 ms 13,286 ms 28,969 ms 15,791 ms 3,681 ms

输入 193 193 193 10 166

输出 2,603 2,097 4,365 2,617 14

缓存创建

5m: 0 / 1h: 0

5,979

5m: 0 / 1h: 5,979

缓存读取 0 0 0 0 0

输出速度 205.3 tok/s 179.2 tok/s 169.5 tok/s 203.7 tok/s -

响应

```json
{"n":72,"c":[8,4,7]}
```

```json
{"n": 72, "c": [8, 4, 7]}
```

```json
{"n": 72, "c": [8, 4, 7]}
```

Looking at the string `607467364493081224916910171690428524148292534530601440814950871454645956`, I'll count the digits:

**Length:** 72 characters

**Counting 0s, 7s, and 9s:**
- **0s**: positions 2,

{"n":66,"c":[1,4,3]}

总计

总输入(含缓存) 142,981 143,617 144,002 107,416 44,784

总输出 5,413 6,498 7,769 6,686 223

总缓存创建

42,660

5m: 0 / 1h: 42,660

43,245

5m: 0 / 1h: 43,245

43,609

5m: 0 / 1h: 43,609

39,389

5m: 0 / 1h: 39,389

总缓存读取 100,078 100,129 100,150 67,967 0

输出速度 169.2 tok/s 175.4 tok/s 165.8 tok/s 175.0 tok/s -

总执行时间 52.6s 1m 0.9s 1m 12.4s 1m 1.6s 35.0s

总墙钟时间

8m 51.0s

含步骤间等待

14m 14.2s

含步骤间等待

12m 53.1s

含步骤间等待

14m 54.0s

含步骤间等待

15m 6.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0906

Anthropic: Claude Haiku 4.5

$0.0968

Anthropic: Claude Haiku 4.5

$0.1036

Anthropic: Claude Haiku 4.5

$0.0895

Anthropic: Claude Haiku 4.5

$0.0459

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%) direct (100%) direct (100%) direct (50%)

上游 Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) -

CDN - cloudflare - - -

ID 格式 msg_01QfckMY67FxQDimAZKSfkuB msg_018QPY3ZGztrqLjB3rX46j3n msg_01KpL8m9TqJad76FQcR6HxoX msg_01KLqDHD4Eh2abmSpJxAFnQg 自定义

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_011CcKJFXRzGwSaFxKR1WMEv
M-	cf-ray: a0ffa29adef3734b-NRT server: cloudflare set-cookie: server_name_session=46db03c0040e9983379bcced7435c21b; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 20260623011419328685555ZyeX4ABd	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	a0ffa29adef3734b-NRT
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: e57601a4518a5d6fdab40786b99b2116-IAD server: nginx	id_format: msg_prefix via_chain: ["1.1 google"] extra_fields: ["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_e57601a4518a5d6fdab40786b99b2116 e57601a4518a5d6fdab40786b99b2116-IAD
M-	server: nginx	id_format: non_standard has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	body:id:non_standardbody:model:date_suffixbody:usage.cache_fields	—

协议指纹评分（机器自动）

总分 / 100 100 98 98 93 37

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 10.0 /10 ×20 10.0 /10 ×20 0.0 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 0.0 /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 5.0 /10 ×14

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

原生 msg-ID 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 0.0 /10 ×8

系统提示纯净 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 0.0 /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 5.0 /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 0.0 /10 ×6

inference_geo 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 0.0 /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 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

与基线相似度基准（参考） 98 v3.24.0 98 v3.24.0 93 v3.24.0 37 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 7716 ms 8998 ms 10691 ms 9370 ms 3873 ms

输出速度 169.2 tok/s 175.4 tok/s 165.8 tok/s 175.0 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 ⓘ

$0.0906

$0.0968

$0.1036

$0.0895

$0.0459

缓存读取占比 ⓘ 70% 70% 70% 63% 0%

揭晓

通道

官方基准

服务商 LingxiCode ↗

通道M-

服务商 InfAI ↗

通道R-Kiro

服务商 ClaudeZZ ↗

通道R-kiro

服务商 AnyAIGC ↗

通道M-

返回列表