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

2026/07/01 运行日期

claude-sonnet-4-6 官方基准

2 通道分数 < 70 另含 1 任务失败

已揭晓揭晓进度

步骤 / 指标

官方基准

NowCoding

R-Kiro

AiTokensFlux

O-Pro

SSSAiCode

O-Max

WorldBase.ai

M-

基本信息

服务名 Claude Code Claude Code Claude Code Claude Code Claude Code

模型 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6

客户端 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195

偏差标红基于官方基准

步骤 1: ping text

时间 17:33:29 ~ 17:33:44 (11.6s) 17:33:21 ~ 17:33:26 (4.7s) 17:33:25 ~ 17:33:30 (4.2s) 17:33:23 ~ 17:33:28 (3.5s) 17:33:27 ~ 17:33:32 (4.9s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 10,331 ms 3,590 ms 3,079 ms 2,198 ms 3,335 ms

首文本(SSE delta) 10,390 ms 3,591 ms 3,095 ms 2,285 ms 3,486 ms

总耗时 10,589 ms 3,610 ms 3,349 ms 2,583 ms 3,510 ms

输入 3 21,416 3 3 4

输出 5 2 5 5 1

缓存创建

5,991

5m: 0 / 1h: 5,991

20,875

5m: 0 / 1h: 20,875

20,517

5m: 0 / 1h: 20,517

20,624

5m: 0 / 1h: 20,624

缓存读取 14,884 0 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 17:36:12 ~ 17:36:27 (10.4s) 17:37:09 ~ 17:37:14 (3.3s) 17:36:22 ~ 17:36:27 (3.9s) 17:36:05 ~ 17:36:10 (4.0s) 17:35:53 ~ 17:36:03 (8.9s)

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。自报身份，输出严格遵守以下 3 行格式（不允许 markdown、代码块、额外文本）：
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) 8,461 ms 2,281 ms 2,573 ms 2,978 ms 7,298 ms

首文本(SSE delta) 8,463 ms 2,282 ms 2,574 ms 3,080 ms 7,319 ms

总耗时 9,406 ms 2,302 ms 3,186 ms 3,259 ms 7,741 ms

输入 3 21,632 3 3 12

输出 22 16 22 22 14

缓存创建

192

5m: 0 / 1h: 192

21,067

5m: 0 / 1h: 21,067

192

5m: 0 / 1h: 192

129

5m: 0 / 1h: 129

缓存读取 20,875 0 0 20,517 20,624

响应

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

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

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

步骤 3: cutoff text

时间 17:39:00 ~ 17:39:20 (19.4s) 17:38:58 ~ 17:39:08 (5.3s) 17:39:04 ~ 17:39:09 (3.7s) 17:39:02 ~ 17:39:07 (3.3s) 17:38:01 ~ 17:38:11 (6.9s)

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) 18,158 ms 3,761 ms 2,695 ms 2,269 ms 5,607 ms

首文本(SSE delta) 18,354 ms 3,763 ms 2,697 ms 2,405 ms 5,746 ms

总耗时 18,458 ms 4,264 ms 2,885 ms 2,499 ms 5,772 ms

输入 3 43 3 3 20

输出 8 22 8 8 2

缓存创建

5m: 0 / 1h: 55

21,611

5m: 21,611 / 1h: 0

5m: 0 / 1h: 38

5m: 0 / 1h: 55

5m: 0 / 1h: 36

缓存读取 21,067 0 21,067 20,709 20,753

响应

2025-08

2025-08

2025-08

2025-04

2025-08

步骤 4: identity_free text

时间 17:42:37 ~ 17:42:52 (10.0s) 17:40:36 ~ 17:40:46 (6.5s) 17:42:26 ~ 17:42:36 (9.8s) 17:40:29 ~ 17:40:34 (3.6s) 17:39:26 ~ 17:39:36 (6.7s)

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) 8,328 ms 4,435 ms 7,822 ms 2,056 ms 4,266 ms

首文本(SSE delta) 8,333 ms 4,436 ms 7,823 ms 2,655 ms 4,458 ms

总耗时 9,023 ms 5,473 ms 8,955 ms 2,793 ms 5,597 ms

输入 3 43 3 3 28

输出 46 95 57 50 30

缓存创建

5m: 0 / 1h: 41

367

5m: 367 / 1h: 0

5m: 0 / 1h: 35

5m: 0 / 1h: 43

5m: 0 / 1h: 21

缓存读取 21,122 21,611 21,105 20,764 20,789

输出速度 66.7 tok/s 91.6 tok/s 50.4 tok/s - 26.3 tok/s

响应

我叫 Claude，是由 Anthropic 开发的 AI 助手，当前使用的模型版本为 claude-sonnet-4-6。

我叫 Claude，是 Anthropic 开发的 AI 助手，属于 Claude 模型系列（当前版本为 claude-sonnet-4-6）。

<thinking>
Short and direct answer, within 50 characters.
</thinking>

我是 Claude，由 Anthropic 开发的 AI 助手。我基于大型语言模型构建，旨在提供安全、有帮助的对话和任务协助。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助回答问题、分析代码、撰写文本等多种任务。

我是 Kiro，一个由 Anthropic 的 Claude 模型驱动的 AI 开发环境助手，专为软件工程任务设计。

步骤 5: knowledge_recall text

时间 17:44:48 ~ 17:45:03 (11.8s) 17:42:46 ~ 17:43:21 (30.0s) 17:46:03 ~ 17:46:08 (4.7s) 17:41:59 ~ 17:42:09 (5.0s) 17:42:21 ~ 17:42:31 (5.9s)

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) 9,066 ms 7,339 ms 2,102 ms 1,528 ms 2,747 ms

首文本(SSE delta) 9,095 ms 7,341 ms 2,108 ms 3,746 ms 3,007 ms

总耗时 10,746 ms 29,043 ms 3,968 ms 4,280 ms 4,931 ms

输入 3 43 3 3 36

输出 205 1,841 204 198 142

缓存创建

6,666

5m: 0 / 1h: 6,666

9,778

5m: 9,778 / 1h: 0

8,785

5m: 0 / 1h: 8,785

8,445

5m: 0 / 1h: 8,445

9,055

5m: 0 / 1h: 9,055

缓存读取 2,131 0 0 0 0

输出速度 124.2 tok/s 84.8 tok/s 109.7 tok/s burst 73.8 tok/s

响应

{"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":"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":"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":"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":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"Real Madrid","nba_2024":"Boston Celtics","copa_america_2024":"Argentina","euro2024":"Spain","nobel

步骤 6: digit_count text

时间 17:46:18 ~ 17:46:38 (16.2s) 17:44:42 ~ 17:45:22 (39.0s) 17:47:54 ~ 17:48:14 (19.5s) 17:45:10 ~ 17:45:15 (1.1s) 17:44:58 ~ 17:45:13 (14.4s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=759858622223461407374738329236679215827948887848297161352099768561400063

HTTP 头(TTFB) 1,423 ms 20,322 ms 3,319 ms - 2,565 ms

首文本(SSE delta) 1,824 ms 20,324 ms 3,321 ms - 2,566 ms

总耗时 15,346 ms 38,095 ms 18,800 ms - 13,025 ms

输入 164 66 164 - 145

输出 1,091 1,087 1,405 - 234

缓存创建

5m: 0 / 1h: 0

144

5m: 144 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 0 - 0

输出速度 80.7 tok/s 61.2 tok/s 90.8 tok/s - 22.4 tok/s

响应

## Counting digits in `s`

**s** = `759858622223461407374738329236679215827948887848297161352099768561400063`

### Step 1 — Length

Splitting into groups of 10:
```
7598586222  (10)
2346140737  (10)
4

```json
{"n":72,"c":[5,9,7]}
```

**Counting digits in:**
`759858622223461407374738329236679215827948887848297161352099768561400063`

**Length:** Chunking into groups of 4:
`7598 5862 2223 4614 0737 4738 3292 3667 9215 8279 4888 7848

失败

评测失败

{"n":72,"c":[5,9,7]}

总计

总输入(含缓存)

93,203

96,754

93,151

91,257

仅前 6 步

92,276

总输出

1,377

3,063

1,701

283

仅前 6 步

423

总缓存创建

12,945

5m: 0 / 1h: 12,945

31,900

5m: 31,900 / 1h: 0

50,800

5m: 0 / 1h: 50,800

29,252

5m: 0 / 1h: 29,252

仅前 6 步

29,865

5m: 0 / 1h: 29,865

总缓存读取

80,079

21,611

42,172

61,990

仅前 6 步

62,166

输出速度

84.6 tok/s

74.6 tok/s

90.2 tok/s

burst

仅前 6 步

30.0 tok/s

总执行时间

1m 19.3s

1m 28.8s

45.9s

20.5s

仅前 6 步

47.7s

总墙钟时间

13m 9.0s

含步骤间等待

12m 1.1s

含步骤间等待

14m 49.0s

含步骤间等待

11m 52.0s

含步骤间等待

仅前 6 步

11m 46.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0938

Anthropic: Claude Sonnet 4.6

$0.2569

Anthropic: Claude Sonnet 4.6

$0.2292

Anthropic: Claude Sonnet 4.6

$0.1326

Anthropic: Claude Sonnet 4.6 (部分)

$0.1377

Anthropic: Claude Sonnet 4.6

通道指纹

来源步骤 digit_count digit_count digit_count knowledge_recall digit_count

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

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

CDN - - - - cloudflare

ID 格式 msg_01128hKxB3oj7N2AXpbFKdR9 msg_01duOB4DxTk4Xf2mG946cQnd msg_01K8nbsXGvjsJgLS9ZnW8KXh msg_01DgjoHpQZEaaTymHrmJUb2K msg_a2bf537848f24d1ca5e5337aad83c762

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: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011Ccb8B9d6cB1hkpTb6R6P8
R-Kiro	server: nginx x-new-api-version: v0.0.0 x-oneapi-request-id: 20260701094443871705944UzkxQufl	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: false usage_has_cache_fields: true	hdr:x-new-api-versionhdr:x-oneapi-request-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
O-Pro	server: nginx	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
O-Max	cf-ray: a144754d897cc4d5-LAX server: Photon-Edge	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	1782898920462-jl1ze1lrn1oha req_011Ccb7r2vF1RffGQWXC8d6r a144754d897cc4d5-LAX
M-	cf-ray: a14479b0e8edb8ae-AMS server: cloudflare	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	hdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	0c6b7416-7b37-445e-b620-b3b90b76b699 a14479b0e8edb8ae-AMS

协议指纹评分（机器自动）

总分 / 100 99 65 85 74

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 0.0 /10 ×20 0.0 /10 ×20 — 7.7 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 0.0 /10 ×15 10.0 /10 ×15 — 10.0 /10 ×15

延迟基线 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 2.3 /10 ×13 7.7 /10 ×13 — 10.0 /10 ×13

原生 msg-ID 10.0 /10 ×8 4.0 /10 ×8 10.0 /10 ×8 — 0.0 /10 ×8

系统提示纯净 10.0 /10 ×8 5.0 /10 ×8 10.0 /10 ×8 — 5.0 /10 ×8

身份(结构) 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 — 0.0 /10 ×7

service_tier 10.0 /10 ×6 4.0 /10 ×6 10.0 /10 ×6 — 0.0 /10 ×6

知识截止 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 — 10.0 /10 ×5

inference_geo 10.0 /10 ×5 4.0 /10 ×5 10.0 /10 ×5 — 0.0 /10 ×5

流式投递 6.9 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 — 10.0 /10 ×5

stop_reason 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3 — 10.0 /10 ×3

Req-ID 透传 10.0 /10 ×2 0.0 /10 ×2 0.0 /10 ×2 — 0.0 /10 ×2

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

综合结论

与基线相似度基准（参考） 65 v3.26.0 85 v3.26.0 执行失败，不评分 74 v3.26.0

雷达图例

共 20 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 流式投递 buffer_dump_match
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 缓存连续 cache_continuity_intra
13 知识召回 knowledge_recall_match
14 延迟基线 latency_baseline_match
15 原生 msg-ID anthropic_msg_id_format
16 Sliding 缓存 cache_sliding_correctness
17 结构化身份 identity_structured_match
18 信封自报一致 envelope_self_report_match
19 档位思考量 tier_thinking_volume_match
20 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 12261 ms 13798 ms 6857 ms 3083 ms 6763 ms

输出速度 84.6 tok/s 74.6 tok/s 90.2 tok/s burst 30.0 tok/s

完成步骤 6/6 成功 6/6 成功 6/6 成功 5/6 成功 · 1 失败 6/6 成功

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0938

$0.2569

$0.2292

$0.1326 (部分)

$0.1377

缓存读取占比 ⓘ 86% 22% 45% 68% (部分) 67%

揭晓

通道

官方基准

服务商 NowCoding ↗

通道R-Kiro

服务商 AiTokensFlux ↗

通道O-Pro

服务商 SSSAiCode ↗

通道O-Max

服务商 WorldBase.ai ↗

通道M-

返回列表