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

2026/06/24 运行日期

claude-sonnet-4-6 官方基准

3 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

SSSAiCode

R-Kiro

FastCode

官方基准

luminai

M-Kiro/Antigravity

OmniaKey

O-Max

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 v2.1.160 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 04:15:05 ~ 04:15:15 (5.7s) 04:15:09 ~ 04:15:14 (4.7s) 04:15:11 ~ 04:15:16 (4.3s) 04:15:07 ~ 04:15:17 (7.0s) 04:15:03 ~ 04:15:13 (6.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,745 ms 3,132 ms 2,926 ms 4,956 ms 4,720 ms

首文本(SSE delta) 3,828 ms 3,134 ms 2,928 ms 4,999 ms 4,728 ms

总耗时 3,894 ms 3,425 ms 3,094 ms 5,074 ms 4,841 ms

输入 0 3 3 1,443 3

输出 2 5 5 1 5

缓存创建

32,318

5m: 0 / 1h: 32,318

32,318

5m: 0 / 1h: 32,318

4,111

5m: 4,111 / 1h: 0

32,318

5m: 0 / 1h: 32,318

缓存读取 0 0 0 23,298 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 04:18:36 ~ 04:18:41 (3.9s) 04:18:12 ~ 04:18:22 (5.1s) 04:18:44 ~ 04:18:49 (3.2s) 04:18:07 ~ 04:18:17 (6.3s) 04:18:14 ~ 04:18:24 (7.8s)

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) 2,711 ms 2,959 ms 1,869 ms 5,003 ms 3,935 ms

首文本(SSE delta) 2,726 ms 2,960 ms 1,870 ms 5,083 ms 3,936 ms

总耗时 2,948 ms 3,537 ms 2,219 ms 5,307 ms 6,571 ms

输入 0 3 3 1,452 3

输出 17 22 22 24 22

缓存创建

32,509

5m: 0 / 1h: 32,509

191

5m: 0 / 1h: 191

4,137

5m: 4,137 / 1h: 0

191

5m: 0 / 1h: 191

缓存读取 0 0 32,318 23,445 32,318

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-4-6

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

步骤 3: cutoff text

时间 04:22:09 ~ 04:22:14 (3.6s) 04:21:02 ~ 04:21:12 (4.9s) 04:21:58 ~ 04:22:03 (3.1s) 04:20:56 ~ 04:21:06 (6.5s) 04:21:15 ~ 04:21:25 (6.1s)

提示词

查看本步 Prompt

请直接回答你的知识截止日期，格式 YYYY-MM，不要附加任何说明。

HTTP 头(TTFB) 2,619 ms 3,202 ms 1,937 ms 5,034 ms 4,360 ms

首文本(SSE delta) 2,798 ms 3,204 ms 1,938 ms 5,103 ms 4,611 ms

总耗时 2,804 ms 3,454 ms 2,109 ms 5,209 ms 4,770 ms

输入 0 3 3 1,455 3

输出 3 8 8 4 8

缓存创建

32,547

5m: 0 / 1h: 32,547

5m: 0 / 1h: 38

4,145

5m: 4,145 / 1h: 0

5m: 0 / 1h: 38

缓存读取 0 0 32,509 23,494 32,509

响应

2025-08

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 04:25:12 ~ 04:25:17 (4.8s) 04:23:26 ~ 04:23:31 (4.8s) 04:25:06 ~ 04:25:11 (3.9s) 04:23:37 ~ 04:23:47 (7.2s) 04:25:24 ~ 04:25:34 (6.1s)

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) 2,967 ms 2,863 ms 1,958 ms 5,751 ms 3,942 ms

首文本(SSE delta) 3,123 ms 2,864 ms 1,959 ms 6,027 ms 4,422 ms

总耗时 3,965 ms 3,572 ms 2,910 ms 6,082 ms 4,796 ms

输入 0 3 3 1,457 3

输出 30 50 64 24 46

缓存创建

32,590

5m: 0 / 1h: 32,590

5m: 0 / 1h: 43

4,151

5m: 4,151 / 1h: 0

5m: 0 / 1h: 39

缓存读取 0 0 32,547 23,523 32,547

输出速度 35.6 tok/s 70.6 tok/s 67.3 tok/s - 123.0 tok/s

响应

我是 Kiro，一个由 Anthropic 开发的 AI 开发环境，基于 Claude 模型构建，专注于帮助开发者完成软件工程任务。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助回答问题、分析代码、写作、推理等多种任务。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助编写代码、回答问题、分析文本等多种任务。当前版本为 claude-sonnet-4-6。

我是 Claude Sonnet 4.6，由 Anthropic 开发的 AI 助手。

我是 Claude，由 Anthropic 公司打造的 AI 助手。当前运行的模型版本为 claude-sonnet-4-6。

步骤 5: knowledge_recall text

时间 04:27:02 ~ 04:27:12 (4.9s) 04:25:21 ~ 04:25:31 (5.6s) 04:26:40 ~ 04:26:50 (5.4s) 04:26:59 ~ 04:27:09 (8.0s) 04:29:17 ~ 04:29:27 (5.3s)

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) 1,878 ms 2,011 ms 1,905 ms 3,804 ms 2,032 ms

首文本(SSE delta) 1,910 ms 2,013 ms 1,907 ms 6,213 ms 2,033 ms

总耗时 3,934 ms 4,295 ms 4,415 ms 6,357 ms 3,931 ms

输入 0 3 3 373 3

输出 221 204 205 210 204

缓存创建

8,029

5m: 0 / 1h: 8,029

8,043

5m: 0 / 1h: 8,043

1,062

5m: 1,062 / 1h: 0

8,021

5m: 0 / 1h: 8,021

缓存读取 0 0 0 6,024 0

输出速度 109.2 tok/s 89.4 tok/s 81.7 tok/s burst 107.5 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 04:28:42 ~ 04:29:07 (20.4s) 04:29:11 ~ 04:29:31 (18.9s) 04:28:33 ~ 04:28:48 (14.4s) 04:30:46 ~ 04:30:57 (5.0s) 04:31:54 ~ 04:32:09 (14.3s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=494837379530357273809865406306244344847808389782706436486627850782834745

HTTP 头(TTFB) 1,929 ms 1,565 ms 1,082 ms 3,437 ms 3,122 ms

首文本(SSE delta) 1,974 ms 1,884 ms 1,086 ms 3,882 ms 3,123 ms

总耗时 19,401 ms 17,756 ms 13,474 ms 3,934 ms 13,164 ms

输入 0 164 164 191 164

输出 1,323 1,462 1,290 14 871

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 - 0

输出速度 75.9 tok/s 92.1 tok/s 104.1 tok/s - 86.7 tok/s

响应

```
s = "494837379530357273809865406306244344847808389782706436486627850782834745"
```

Mapping each position (1-indexed):

| pos | digit | pos | digit | pos | digit | pos
|-----|-------|-----|-------

I'll count each character in:
`494837379530357273809865406306244344847808389782706436486627850782834745`

**Indexing each digit (1-based):**

| pos | d | pos | d | pos | d | pos | d |
|-----|---|-----

## Counting digits in `s`

**String:** `494837379530357273809865406306244344847808389782706436486627850782834745`

### Length

Going through the string character by character:

`4 9 4 8 3 7 3 7 9 5 3

{"n":71,"c":[4,5,3]}

```
s = "494837379530357273809865406306244344847808389782706436486627850782834745"
```

**Step-by-step count:**

| Digit | Positions (1-indexed) | Count |
|---|---|---|
| **0** | 12, 20, 26, 29, 41, 5

总计

总输入(含缓存) 0 138,172 138,186 123,761 138,160

总输出 1,596 1,751 1,594 277 1,156

总缓存创建

137,993

5m: 0 / 1h: 137,993

40,633

5m: 0 / 1h: 40,633

17,606

5m: 17,606 / 1h: 0

40,607

5m: 0 / 1h: 40,607

总缓存读取 0 0 97,374 99,784 97,374

输出速度 77.6 tok/s 91.0 tok/s 98.4 tok/s burst 91.0 tok/s

总执行时间 43.3s 43.9s 34.4s 40.0s 45.8s

总墙钟时间

14m 2.1s

含步骤间等待

14m 21.9s

含步骤间等待

13m 37.0s

含步骤间等待

15m 49.1s

含步骤间等待

17m 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 ⓘ

>=$0.0239

Anthropic: Claude Sonnet 4.6 (成本不完整)

$0.5443

Anthropic: Claude Sonnet 4.6

$0.2060

Anthropic: Claude Sonnet 4.6

$0.0945

Anthropic: Claude Sonnet 4.6

$0.1994

Anthropic: Claude Sonnet 4.6

通道指纹

来源步骤 digit_count digit_count digit_count digit_count digit_count

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

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

CDN - - - cloudflare cloudflare

ID 格式 msg_eab6d2c7daf7400783d3c9b0bcda49c9 msg_vrtx_01EcsvRCNem8xSvuzFgd2Fbz msg_01HAFeNCzM3947NSsc2x7NU6 msg_016fUvYvFJTCsdiCDbW2GJT9Vn msg_01ExExp3ABCMC35S5WnPZBxy

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: Photon-Edge x-amzn-requestid: d44e8809-e065-41ec-be0f-d8958e51e500	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	1782246524084-1hheztk169pi7 d44e8809-e065-41ec-be0f-d8958e51e500
cc	server: nginx	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	body:id:msg_prefixbody: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: 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_011CcLpgQetjFAMxbYLiZxtF
M-Kiro/Antigravity	cf-ray: a10640aefdcc3c32-AMS server: cloudflare x-oneapi-request-id: 202606232030486292767338268d9d6WcFRlaGc	id_format: msg_prefix extra_fields: ["container"] has_service_tier: true has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	hdr:x-oneapi-request-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a10640aefdcc3c32-AMS
O-Max	cf-ray: a10642575f1d4a19-AMS server: cloudflare x-oneapi-request-id: 202606232031565420692398268d9d6PAHET92L	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-oneapi-request-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a10642575f1d4a19-AMS

协议指纹评分（机器自动）

总分 / 100 52 68 100 73 98

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 0.0 /10 ×20 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 0.0 /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 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 10.0 /10 ×14

sliding 5m 5.0 /10 ×13 0.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13

原生 msg-ID 0.0 /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 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 10.0 /10 ×7

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

service_tier 0.0 /10 ×6 0.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

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

延迟基线 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 7.4 /10 ×5 9.6 /10 ×5

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

Req-ID 透传 0.0 /10 ×4 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 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

系统提示纯净未计入 10.0 /10 ×8 10.0 /10 ×8 0.0 /10 ×8 10.0 /10 ×8

综合结论

与基线相似度 52 v3.24.0 68 v3.24.0 基准（参考） 73 v3.24.0 98 v3.24.0

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 Service Tier service_tier_present
8 缓存命中比 cache_hit_ratio_match
9 缓存 TTL 一致 cache_ttl_consistency
10 推理地区 inference_geo_present
11 缓存连续 cache_continuity_intra
12 知识召回 knowledge_recall_match
13 延迟基线 latency_baseline_match
14 原生 msg-ID anthropic_msg_id_format
15 Sliding 缓存 cache_sliding_correctness
16 结构化身份 identity_structured_match
17 信封自报一致 envelope_self_report_match
18 中文遵循 instruction_following_lang
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 6158 ms 6007 ms 4704 ms 5327 ms 6346 ms

输出速度 77.6 tok/s 91.0 tok/s 98.4 tok/s burst 91.0 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 ⓘ

>=$0.0239 (成本不完整)

$0.5443

$0.2060

$0.0945

$0.1994

缓存读取占比 ⓘ - 0% 70% 81% 70%

揭晓

通道

服务商 SSSAiCode ↗

通道R-Kiro

服务商 FastCode ↗

通道cc

官方基准

服务商 luminai ↗

通道M-Kiro/Antigravity

服务商 OmniaKey ↗

通道O-Max

返回列表