模型对比 | RPDiag

测试说明点击展开 5 步 quick-probe-v1 详解

同一会话连续 5 步，覆盖 5 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。步与步之间随机延迟 1–4 分钟。第 3 步（cutoff）累计跨度 > 5 分钟保住 sliding cache 检测；全流程总跨度 > 6 分钟容纳新增的第 5 步。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。

3 通道 · 5 步骤

2026/06/15 运行日期

claude-opus-4-8 官方基准

2 任务失败执行不完整

已揭晓揭晓进度

⚠ 2 通道不可测 · 已从主对比表移除点击展开原因

TopRouterCN / O-Max
```
评测失败
```
DBAI / M-
```
评测失败
```

步骤 / 指标

官方基准

WorldBase.ai

M-

ClaudeZZ

R-kiro

基本信息

服务名 Claude Code Claude Code Claude Code

模型 claude-opus-4-8 claude-opus-4-8 claude-opus-4-8

客户端 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 20:27:40 ~ 20:27:45 (3.1s) 20:27:34 ~ 20:27:44 (5.6s) 20:27:36 ~ 20:27:46 (8.0s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 1,874 ms 2,988 ms 6,284 ms

首文本(SSE delta) 1,876 ms 2,995 ms 6,288 ms

总耗时 2,005 ms 4,337 ms 6,828 ms

输入 1,454 1,152 2

输出 4 66 4

缓存创建

30,222

5m: 0 / 1h: 30,222

23,169

5m: 0 / 1h: 23,169

27,914

5m: 0 / 1h: 27,914

缓存读取 0 0 0

输出速度 - 49.2 tok/s -

响应

pong

I can't discuss that.

It looks like there's no actual question or task here yet — just the system setup and a list of available skills.

What would you like to work on? I'm in `/playground` and ready

pong

步骤 2: identity text

时间 20:29:37 ~ 20:29:42 (3.1s) 20:29:59 ~ 20:30:14 (14.1s) 20:31:07 ~ 20:31:17 (7.3s)

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。请只输出以下 3 行内容（label 小写，值小写）：
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) 1,754 ms 11,073 ms 6,121 ms

首文本(SSE delta) 1,756 ms 11,074 ms 6,131 ms

总耗时 2,160 ms 12,270 ms 6,485 ms

输入 2 197 2

输出 28 20 28

缓存创建

1,630

5m: 0 / 1h: 1,630

1,232

5m: 0 / 1h: 1,232

190

5m: 0 / 1h: 190

缓存读取 30,222 23,165 27,914

响应

vendor: anthropic
brand: claude
model: claude-opus-4-8

vendor: anthropic
brand: claude
model: claude-opus-4-8

vendor: anthropic
brand: claude
model: claude-opus-4-8

步骤 3: cutoff text

时间 20:31:49 ~ 20:31:54 (3.5s) 20:31:55 ~ 20:32:00 (4.9s) 20:33:42 ~ 20:33:52 (7.2s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 2,101 ms 2,787 ms 5,735 ms

首文本(SSE delta) 2,102 ms 2,788 ms 5,737 ms

总耗时 2,326 ms 3,638 ms 6,458 ms

输入 2 37 2

输出 6 4 6

缓存创建

5m: 0 / 1h: 54

220

5m: 0 / 1h: 220

100

5m: 0 / 1h: 100

缓存读取 31,852 24,393 28,104

响应

2026-01

2026-01

2025-02

步骤 4: identity_free text

时间 20:34:31 ~ 20:34:36 (3.8s) 20:33:21 ~ 20:33:31 (6.6s) 20:37:25 ~ 20:37:35 (5.8s)

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) 2,177 ms 2,890 ms 3,977 ms

首文本(SSE delta) 2,179 ms 2,891 ms 4,278 ms

总耗时 2,832 ms 5,068 ms 5,011 ms

输入 2 29 2

输出 56 55 53

缓存创建

5m: 0 / 1h: 37

5m: 0 / 1h: 40

5m: 0 / 1h: 54

缓存读取 31,906 24,603 28,204

输出速度 85.8 tok/s 25.3 tok/s 72.3 tok/s

响应

我是 Claude，由 Anthropic 开发的 AI 助手，具体型号为 Claude Opus 4.8，属于 Claude 系列产品。

我是 Claude，一个由 Anthropic 开发的 AI 助手。在这里我以 Kiro 开发环境的形式运行，帮你处理编程、研究、调试等各类任务。

我是 Claude Code，Anthropic 开发的官方命令行编程助手，基于 Claude 模型，可帮助完成软件工程任务。

步骤 5: knowledge_recall text

时间 20:36:53 ~ 20:37:03 (8.0s) 20:36:57 ~ 20:37:17 (15.5s) 20:40:20 ~ 20:40:35 (13.7s)

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) 2,191 ms 9,498 ms 10,764 ms

首文本(SSE delta) 4,546 ms 9,500 ms 10,767 ms

总耗时 7,024 ms 13,299 ms 12,908 ms

输入 2 961 2

输出 692 197 826

缓存创建

4,753

5m: 0 / 1h: 4,753

3,095

5m: 0 / 1h: 3,095

8,044

5m: 0 / 1h: 8,044

缓存读取 0 0 3,067

输出速度 burst 51.9 tok/s burst

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

总计

总输入(含缓存) 132,138 102,293 123,601

总输出 786 342 917

总缓存创建

36,696

5m: 0 / 1h: 36,696

27,756

5m: 0 / 1h: 27,756

36,302

5m: 0 / 1h: 36,302

总缓存读取 93,980 72,161 87,289

输出速度 85.8 tok/s 43.5 tok/s 72.3 tok/s

总执行时间 21.5s 46.6s 41.9s

总墙钟时间

9m 22.9s

含步骤间等待

9m 42.8s

含步骤间等待

12m 58.9s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.3033

Anthropic: Claude Opus 4.8

$0.2300

Anthropic: Claude Opus 4.8

$0.2935

Anthropic: Claude Opus 4.8

通道指纹

来源步骤 knowledge_recall knowledge_recall knowledge_recall

平台 direct (100%) direct (100%) direct (100%)

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

CDN - cloudflare -

ID 格式 msg_01Jd6J3x1YViaCjwXi3DUK1U msg_01MYjpnsmiN2vLAbaWJwBYuK msg_014TfPgyh443oFVqpKu2Jgc3

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_011Cc54ExNTapWdj7JS2sTUt

M-

cf-ray: a0c19fa089b57e47-AMS
server: cloudflare

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:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body

2f0bb3c7-10e2-4497-9f8a-ecad2d80baef
a0c19fa089b57e47-AMS

R-kiro

via: 1.1 google
cf-ray: ec23d5cb2a480bd66afe5b7429d5b201-IAD
server: nginx

id_format: msg_prefix
via_chain: ["1.1 google"]
extra_fields: ["diagnostics","stop_details"]
has_service_tier: true
has_inference_geo: true
model_has_date_suffix: false
usage_has_cache_fields: true

hdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body

req_ec23d5cb2a480bd66afe5b7429d5b201
ec23d5cb2a480bd66afe5b7429d5b201-IAD

协议指纹评分（机器自动）

总分 / 100 96 88 90

基线状态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

缓存 TTL 一致 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

缓存连续性 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

sliding 5m 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

系统提示纯净 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

知识截止 0.0 /10 ×7 0.0 /10 ×7 10.0 /10 ×7

身份(自由) 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

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

延迟基线 10.0 /10 ×5 10.0 /10 ×5 1.5 /10 ×5

Req-ID 透传 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

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

流式投递 — — —

综合结论

与基线相似度基准（参考） 88 v3.16.0 90 v3.16.0

雷达图例

共 19 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 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 中文遵循 instruction_following_lang
19 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 3269 ms 7722 ms 7538 ms

输出速度 85.8 tok/s 43.5 tok/s 72.3 tok/s

完成步骤 5/5 成功 5/5 成功 5/5 成功

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.3033

$0.2300

$0.2935

缓存读取占比 ⓘ 71% 71% 71%

揭晓

通道

官方基准

服务商 WorldBase.ai ↗

通道M-

服务商 ClaudeZZ ↗

通道R-kiro

返回列表