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

2026/06/19 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

AnyAIGC

M-

DawAPI

InfAI

R-Kiro

0-0

O-Max

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 05:40:29 ~ 05:40:34 (3.8s) 05:40:23 ~ 05:40:28 (4.8s) 05:40:25 ~ 05:40:35 (5.9s) 05:40:21 ~ 05:40:26 (3.9s) 05:40:27 ~ 05:40:37 (5.7s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,369 ms 3,154 ms 3,831 ms 1,989 ms 3,022 ms

首文本(SSE delta) 2,370 ms 3,192 ms 3,833 ms 1,992 ms 3,023 ms

总耗时 2,750 ms 3,313 ms 4,938 ms 2,407 ms 3,973 ms

输入 10 8,600 10 10 10

输出 53 2 51 53 54

缓存创建

33,122

5m: 0 / 1h: 33,122

33,122

5m: 33,122 / 1h: 0

33,122

5m: 0 / 1h: 33,122

33,192

5m: 33,192 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 139.5 tok/s - 46.2 tok/s 127.7 tok/s 56.8 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 05:43:00 ~ 05:43:05 (4.5s) 05:43:14 ~ 05:43:19 (4.9s) 05:44:34 ~ 05:44:44 (5.1s) 05:43:18 ~ 05:43:28 (5.1s) 05:43:51 ~ 05:44:01 (6.8s)

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) 1,871 ms 3,540 ms 2,558 ms 1,723 ms 2,737 ms

首文本(SSE delta) 1,873 ms 3,542 ms 2,559 ms 1,725 ms 2,739 ms

总耗时 3,447 ms 3,665 ms 4,325 ms 3,716 ms 5,807 ms

输入 10 8,795 10 10 10

输出 184 28 188 282 232

缓存创建

244

5m: 0 / 1h: 244

242

5m: 242 / 1h: 0

244

5m: 0 / 1h: 244

189

5m: 189 / 1h: 0

缓存读取 33,122 0 33,122 33,122 33,192

输出速度 116.9 tok/s - 106.5 tok/s 141.6 tok/s 75.6 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 05:46:28 ~ 05:46:33 (3.2s) 05:46:20 ~ 05:46:30 (7.9s) 05:48:28 ~ 05:48:33 (3.8s) 05:44:48 ~ 05:44:53 (3.7s) 05:47:11 ~ 05:47:21 (5.0s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 1,756 ms 6,688 ms 2,499 ms 1,943 ms 2,904 ms

首文本(SSE delta) 1,758 ms 6,689 ms 2,500 ms 1,944 ms 2,905 ms

总耗时 2,189 ms 6,753 ms 3,003 ms 2,412 ms 3,767 ms

输入 10 8,858 10 10 10

输出 64 5 78 65 67

缓存创建

207

5m: 0 / 1h: 207

228

5m: 228 / 1h: 0

322

5m: 0 / 1h: 322

5m: 42 / 1h: 0

缓存读取 33,366 0 33,364 33,366 33,381

输出速度 148.5 tok/s - 155.1 tok/s 138.9 tok/s 77.7 tok/s

响应

2025-02

2025-05

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 05:50:31 ~ 05:50:41 (7.5s) 05:50:13 ~ 05:50:18 (4.7s) 05:51:53 ~ 05:52:08 (11.4s) 05:47:57 ~ 05:48:07 (6.6s) 05:51:08 ~ 05:51:18 (7.5s)

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) 2,026 ms 3,296 ms 2,307 ms 3,087 ms 3,239 ms

首文本(SSE delta) 2,028 ms 3,297 ms 2,309 ms 3,089 ms 3,240 ms

总耗时 6,512 ms 3,395 ms 10,649 ms 5,357 ms 6,157 ms

输入 10 8,894 10 10 10

输出 569 21 1,138 182 230

缓存创建

104

5m: 0 / 1h: 104

112

5m: 112 / 1h: 0

5m: 0 / 1h: 99

5m: 35 / 1h: 0

缓存读取 33,573 0 33,592 33,688 33,423

输出速度 126.9 tok/s - 136.5 tok/s 80.2 tok/s 78.8 tok/s

响应

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

我叫Claude，属于Anthropic产品，由Anthropic团队开发，用于协助软件工程。

我是Claude，由Anthropic公司开发的AI助手。可帮助用户完成软件工程任务，如代码编写、调试、测试等。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我能帮助用户完成各种软件工程任务、代码编辑和问题解答。

我是Claude，由Anthropic公司开发的AI助手。我的当前版本是Claude Haiku 4.5，专长是帮助用户完成编程和软件工程任务。

步骤 5: knowledge_recall text

时间 05:53:48 ~ 05:54:08 (19.0s) 05:53:30 ~ 05:53:35 (4.0s) 05:55:33 ~ 05:56:08 (31.0s) 05:50:49 ~ 05:51:09 (15.8s) 05:52:46 ~ 05:53:26 (35.7s)

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) 1,515 ms 2,286 ms 1,940 ms 1,428 ms 2,514 ms

首文本(SSE delta) 1,517 ms 2,287 ms 1,941 ms 1,429 ms 2,515 ms

总耗时 17,993 ms 2,803 ms 30,202 ms 14,584 ms 34,347 ms

输入 10 9,514 10 10 10

输出 2,745 154 4,063 1,979 2,863

缓存创建

8,864

5m: 0 / 1h: 8,864

9,460

5m: 0 / 1h: 9,460

8,587

5m: 0 / 1h: 8,587

8,056

5m: 8,056 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 166.6 tok/s burst 143.8 tok/s 150.4 tok/s 89.9 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

{"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

```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 05:56:42 ~ 05:57:27 (44.1s) 05:56:55 ~ 05:57:00 (4.7s) 05:59:05 ~ 06:00:25 (76.8s) 05:53:08 ~ 05:53:33 (23.1s) 05:55:52 ~ 05:56:17 (22.8s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=729713443517888221641862840522383639376242996087143491428994756857878071

HTTP 头(TTFB) 1,475 ms 3,216 ms 1,133 ms 809 ms 2,576 ms

首文本(SSE delta) 1,557 ms 3,217 ms 1,134 ms 811 ms 2,577 ms

总耗时 43,120 ms 3,448 ms 15,499 ms 21,883 ms 21,420 ms

输入 193 166 193 193 193

输出 4,016 14 2,916 4,012 3,572

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 96.6 tok/s - 203.0 tok/s 190.4 tok/s 189.6 tok/s

响应

```json
{"n":72,"c":[3,9,7]}
```

{"n":78,"c":[1,8,3]}

```json
{"n":72,"c":[3,9,7]}
```

```json
{"n": 72, "c": [3, 9, 7]}
```

```json
{"n": 72, "c": [3, 9, 7]}
```

总计

总输入(含缓存) 142,845 44,827 143,485 142,793 141,753

总输出 7,631 224 8,434 6,573 7,018

总缓存创建

42,541

5m: 0 / 1h: 42,541

43,164

5m: 33,704 / 1h: 9,460

42,374

5m: 0 / 1h: 42,374

41,514

5m: 41,514 / 1h: 0

总缓存读取 100,061 0 100,078 100,176 99,996

输出速度 117.6 tok/s burst 155.2 tok/s 167.0 tok/s 120.0 tok/s

总执行时间 1m 21.9s 30.9s 2m 14.0s 58.1s 1m 23.6s

总墙钟时间

16m 58.1s

含步骤间等待

16m 36.9s

含步骤间等待

20m 0.1s

含步骤间等待

13m 12.0s

含步骤间等待

15m 50.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1016

Anthropic: Claude Haiku 4.5

$0.0459

Anthropic: Claude Haiku 4.5

$0.0906

Anthropic: Claude Haiku 4.5

$0.0961

Anthropic: Claude Haiku 4.5

$0.0778

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%) direct (50%) 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%)

CDN - - - - -

ID 格式 msg_01VnEEKf8Vi7LF94XTFcv3HM 自定义 msg_018vLVzd1ZNBsGSisr3GBZf5 msg_013fdktit7fCM3rtQJDB7uPr msg_01PDx46MXXTCJiW9Wx9uJP22

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_011CcBUMfe9RaFz1GoLWNtqv
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	—
cc	server: cdn x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606182200066341950568268d9d65DS4TGOQ	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	—
O-Max	via: 1.1 Caddy server: uvicorn, nginx/1.24.0 (Ubuntu)	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:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	9c217f1e842d6e4815cf22f117fd6b74

协议指纹评分（机器自动）

总分 / 100 99 35 89 97 89

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 0.0 /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 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 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 0.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8

身份(结构) 10.0 /10 ×7 0.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7

知识截止 10.0 /10 ×7 5.0 /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 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） 35 v3.24.0 89 v3.24.0 97 v3.24.0 89 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 12669 ms 3896 ms 11436 ms 8393 ms 12579 ms

输出速度 117.6 tok/s burst 155.2 tok/s 167.0 tok/s 120.0 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 ⓘ

$0.1016

$0.0459

$0.0906

$0.0961

$0.0778

缓存读取占比 ⓘ 70% 0% 70% 70% 71%

揭晓

通道

官方基准

服务商 AnyAIGC ↗

通道M-

服务商 DawAPI ↗

通道cc

服务商 InfAI ↗

通道R-Kiro

服务商 0-0 ↗

通道O-Max

返回列表