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。

4 通道 · 6 步骤

2026/06/29 运行日期

claude-haiku-4-5-20251001 官方基准

1 通道分数 < 70 另含 1 任务失败

已揭晓揭晓进度

⚠ 1 通道不可测 · 已从主对比表移除点击展开原因

DBAI / M-
```
评测失败
```

步骤 / 指标

WorldBase.ai

M-

官方基准

InfAI

R-Kiro

LingxiCode

M-

基本信息

服务名 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

客户端 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 04:45:49 ~ 04:45:59 (6.8s) 04:45:51 ~ 04:45:56 (3.5s) 04:45:45 ~ 04:45:55 (5.4s) 04:45:47 ~ 04:45:52 (4.2s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 4,593 ms 1,899 ms 4,168 ms 2,840 ms

首文本(SSE delta) 4,595 ms 1,901 ms 4,170 ms 2,842 ms

总耗时 5,313 ms 2,244 ms 4,170 ms 3,345 ms

输入 33,020 10 10 10

输出 44 42 62 47

缓存创建

5m: 0 / 1h: 0

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,123

5m: 0 / 1h: 33,123

缓存读取 0 0 0 0

输出速度 61.3 tok/s 122.4 tok/s burst 93.4 tok/s

响应

pong

pong

pong

pong

步骤 2: identity text

时间 04:47:21 ~ 04:47:31 (8.0s) 04:49:21 ~ 04:49:26 (4.8s) 04:48:20 ~ 04:48:30 (6.3s) 04:47:35 ~ 04:47:45 (6.3s)

提示词

查看本步 Prompt

以下是对你自身的 metadata 探针。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。按以下 3 行回答自己的身份，每行一个 label，不要寒暄不要解释：
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) 4,156 ms 1,657 ms 3,152 ms 2,867 ms

首文本(SSE delta) 4,157 ms 1,660 ms 3,154 ms 2,868 ms

总耗时 6,747 ms 3,881 ms 5,002 ms 5,201 ms

输入 33,575 10 10 10

输出 316 200 292 233

缓存创建

5m: 0 / 1h: 0

246

5m: 0 / 1h: 246

266

5m: 0 / 1h: 266

251

5m: 0 / 1h: 251

缓存读取 0 33,122 33,122 33,123

输出速度 122.0 tok/s 90.0 tok/s 158.0 tok/s 99.9 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

步骤 3: cutoff text

时间 04:48:38 ~ 04:48:48 (7.8s) 04:53:03 ~ 04:53:08 (4.1s) 04:51:29 ~ 04:51:34 (4.1s) 04:51:39 ~ 04:51:44 (4.6s)

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) 6,073 ms 2,154 ms 2,973 ms 3,040 ms

首文本(SSE delta) 6,074 ms 2,156 ms 2,974 ms 3,041 ms

总耗时 6,777 ms 3,050 ms 2,974 ms 3,798 ms

输入 34,255 10 10 10

输出 69 62 62 64

缓存创建

5m: 0 / 1h: 0

223

5m: 0 / 1h: 223

315

5m: 0 / 1h: 315

256

5m: 0 / 1h: 256

缓存读取 0 33,368 33,388 33,374

输出速度 98.2 tok/s 69.4 tok/s burst 84.5 tok/s

响应

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 04:52:32 ~ 04:52:47 (12.6s) 04:56:18 ~ 04:56:28 (4.9s) 04:54:33 ~ 04:54:43 (6.5s) 04:55:25 ~ 04:55:40 (13.3s)

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) 7,746 ms 1,837 ms 3,918 ms 3,010 ms

首文本(SSE delta) 7,747 ms 1,838 ms 3,920 ms 3,011 ms

总耗时 11,410 ms 3,971 ms 5,325 ms 12,504 ms

输入 34,674 10 10 10

输出 638 240 439 1,281

缓存创建

5m: 0 / 1h: 0

104

5m: 0 / 1h: 104

5m: 0 / 1h: 96

106

5m: 0 / 1h: 106

缓存读取 0 33,591 33,703 33,630

输出速度 174.2 tok/s 112.5 tok/s burst 134.9 tok/s

响应

我叫Claude，由Anthropic公司开发。属于Claude产品系列，当前版本是Haiku 4.5。我是AI助手，可帮助软件开发、编程、写作、分析等工作。

我是 Claude Haiku 4.5，由 Anthropic 开发的 AI 助手。我可以帮助您进行编程、分析、写作、问题解答等各类任务。

我是Claude，由Anthropic公司开发的AI助手。我可以帮助用户完成各种任务，包括代码编写、分析、写作和问题解答。

我是Claude，由Anthropic公司开发的AI助手。可以帮助用户完成代码编写、文本分析、问题解答等各类任务。知识库截至2025年2月。

步骤 5: knowledge_recall text

时间 04:56:20 ~ 04:56:45 (21.0s) 04:57:55 ~ 04:58:26 (29.9s) 04:55:56 ~ 04:56:21 (22.8s) 04:57:25 ~ 04:57:40 (10.4s)

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) 3,463 ms 1,454 ms 2,552 ms 1,836 ms

首文本(SSE delta) 3,464 ms 1,455 ms 2,554 ms 1,837 ms

总耗时 19,926 ms 28,922 ms 21,610 ms 9,597 ms

输入 10,647 10 10 10

输出 2,893 3,213 3,235 1,355

缓存创建

5m: 0 / 1h: 0

8,553

5m: 0 / 1h: 8,553

8,856

5m: 0 / 1h: 8,856

9,635

5m: 0 / 1h: 9,635

缓存读取 0 0 0 0

输出速度 175.7 tok/s 117.0 tok/s 169.8 tok/s 174.6 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":"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

```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 04:59:18 ~ 04:59:23 (3.1s) 05:00:04 ~ 05:00:49 (43.4s) 04:59:22 ~ 04:59:42 (18.7s) 04:59:20 ~ 04:59:45 (20.2s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=790259036918717300195190564459358957076019534012241346651191066965146551

HTTP 头(TTFB) 1,890 ms 969 ms 1,977 ms 1,702 ms

首文本(SSE delta) 1,891 ms 971 ms 1,978 ms 1,703 ms

总耗时 2,039 ms 42,413 ms 17,618 ms 19,272 ms

输入 145 193 193 193

输出 8 5,042 2,694 3,463

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0

输出速度 - 121.7 tok/s 172.3 tok/s 197.1 tok/s

响应

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

```json
{"n":72,"c":[9,5,10]}
```

```json
{"n":72,"c":[9,5,10]}
```

```json
{"n":72,"c":[9,5,10]}
```

总计

总输入(含缓存) 146,316 142,572 143,111 143,741

总输出 3,968 8,799 6,784 6,443

总缓存创建

5m: 0 / 1h: 0

42,248

5m: 0 / 1h: 42,248

42,655

5m: 0 / 1h: 42,655

43,371

5m: 0 / 1h: 43,371

总缓存读取 0 100,081 100,213 100,127

输出速度 164.1 tok/s 118.1 tok/s 170.2 tok/s 167.7 tok/s

总执行时间 59.4s 1m 30.6s 1m 3.8s 59.0s

总墙钟时间

13m 33.8s

含步骤间等待

14m 58.0s

含步骤间等待

13m 56.8s

含步骤间等待

13m 57.8s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1662

Anthropic: Claude Haiku 4.5

$0.1071

Anthropic: Claude Haiku 4.5

$0.0975

Anthropic: Claude Haiku 4.5

$0.0967

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count digit_count digit_count

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

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

CDN cloudflare - - cloudflare

ID 格式 msg_7371824582ea4ee3ac5752890ab9dab8 msg_019tpWvwsvjffPwnGSrdtRkU msg_01MPywostsEcAcm9mmdCRHJb msg_01MBs8CmH6psFTXEgDfj8WwT

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
M-	cf-ray: a12f9d54a9d1f5b4-AMS server: cloudflare	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	hdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:model:date_suffixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	86fdf107-1c52-4927-bc1f-e55ef3c4f9a3 a12f9d54a9d1f5b4-AMS
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_011CcWL8KqZDTdnq8XjXoHzw
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	—
M-	cf-ray: a12f9d5fce02e393-NRT server: cloudflare set-cookie: server_name_session=3de609b8cf2a7e28d141c3b6c8b41f57; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 20260628205921513043414YZp4Ml6S	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	a12f9d5fce02e393-NRT

协议指纹评分（机器自动）

总分 / 100 62 98 95 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

缓存命中比 0.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20

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

sliding 5m 5.0 /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 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

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 3.5 /10 ×5 10.0 /10 ×5 8.3 /10 ×5 9.8 /10 ×5

流式投递 10.0 /10 ×5 5.3 /10 ×5 0.0 /10 ×5 0.0 /10 ×5

Req-ID 透传 0.0 /10 ×4 10.0 /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

SDK 一致 10.0 /10 ×2 10.0 /10 ×2 10.0 /10 ×2 10.0 /10 ×2

综合结论

与基线相似度 62 v3.25.0 基准（参考） 95 v3.25.0 95 v3.25.0

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 流式投递 buffer_dump_match
6 自由身份 identity_free_clean
7 停止原因 stop_reason_present
8 系统提示纯净 system_prompt_clean
9 Service Tier service_tier_present
10 缓存命中比 cache_hit_ratio_match
11 缓存 TTL 一致 cache_ttl_consistency
12 推理地区 inference_geo_present
13 Thinking 体量 thinking_volume_match
14 缓存连续 cache_continuity_intra
15 知识召回 knowledge_recall_match
16 延迟基线 latency_baseline_match
17 原生 msg-ID anthropic_msg_id_format
18 Sliding 缓存 cache_sliding_correctness
19 结构化身份 identity_structured_match
20 自报身份一致 self_identity_consistency
21 信封自报一致 envelope_self_report_match
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 8702 ms 14080 ms 9450 ms 8953 ms

输出速度 164.1 tok/s 118.1 tok/s 170.2 tok/s 167.7 tok/s

完成步骤 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 ⓘ

$0.1662

$0.1071

$0.0975

$0.0967

缓存读取占比 ⓘ 0% 70% 70% 70%

揭晓

通道

服务商 WorldBase.ai ↗

通道M-

官方基准

服务商 InfAI ↗

通道R-Kiro

服务商 LingxiCode ↗

通道M-

返回列表