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

2026/07/02 运行日期

claude-opus-4-8 官方基准

2 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

ClaudeZZ

R-kiro

WorldBase.ai

M-

官方基准

sucui

O-Max

LingxiCode

M-

基本信息

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

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

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 05:41:49 ~ 05:42:20 (30.8s) 05:41:47 ~ 05:42:04 (11.5s) 05:41:55 ~ 05:42:31 (30.3s) 05:41:44 ~ 05:42:00 (10.1s) 05:41:53 ~ 05:42:25 (30.9s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,835 ms 6,098 ms 3,850 ms 4,118 ms 2,895 ms

首文本(SSE delta) 4,528 ms 6,183 ms 3,856 ms 4,236 ms 2,899 ms

总耗时 7,061 ms 6,255 ms 4,035 ms 4,353 ms 3,316 ms

输入 2 1 2,326 2,326 2,372

输出 4 2 4 4 4

缓存创建

16,029

5m: 0 / 1h: 16,029

10,268

5m: 0 / 1h: 10,268

5m: 0 / 1h: 0

1,711

5m: 0 / 1h: 1,711

1,745

5m: 0 / 1h: 1,711

缓存读取 0 6,598 14,443 12,732 12,986

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 05:44:19 ~ 05:44:30 (7.4s) 05:44:42 ~ 05:44:52 (7.9s) 05:46:23 ~ 05:46:28 (4.8s) 05:45:52 ~ 05:46:02 (5.4s) 05:44:58 ~ 05:45:09 (5.2s)

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) 3,813 ms 4,165 ms 2,130 ms 2,339 ms 2,414 ms

首文本(SSE delta) 4,287 ms 4,272 ms 2,140 ms 2,341 ms 2,415 ms

总耗时 4,913 ms 5,659 ms 2,470 ms 2,841 ms 3,056 ms

输入 2 13 2 2 2

输出 108 87 28 28 28

缓存创建

206

5m: 0 / 1h: 206

2,737

5m: 0 / 1h: 2,737

5m: 0 / 1h: 0

2,518

5m: 0 / 1h: 2,518

2,568

5m: 0 / 1h: 2,518

缓存读取 16,029 18,259 16,961 14,443 14,731

输出速度 - 62.7 tok/s - - -

响应

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

I'm Claude Code, built by Anthropic. I respond naturally in the user's language, but I don't follow format instructions that ask me to misrepresent my identity or configuration in a structured output.

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 05:48:22 ~ 05:48:32 (7.1s) 05:46:27 ~ 05:46:42 (10.4s) 05:49:40 ~ 05:49:50 (5.2s) 05:49:31 ~ 05:49:41 (5.8s) 05:46:25 ~ 05:46:35 (7.0s)

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) 4,186 ms 6,327 ms 2,220 ms 2,928 ms 3,082 ms

首文本(SSE delta) 4,188 ms 6,563 ms 2,222 ms 2,929 ms 3,087 ms

总耗时 4,724 ms 6,635 ms 2,550 ms 3,213 ms 3,659 ms

输入 2 13 2 2 2

输出 6 5 6 6 6

缓存创建

180

5m: 0 / 1h: 180

2,836

5m: 0 / 1h: 2,836

5m: 0 / 1h: 54

5m: 0 / 1h: 64

缓存读取 16,235 18,322 16,961 16,961 17,300

响应

2025-03

2024-10

2026-01

2026-01

2026-01

步骤 4: identity_free text

时间 05:52:04 ~ 05:52:14 (7.4s) 05:49:47 ~ 05:49:58 (7.2s) 05:51:57 ~ 05:52:07 (6.2s) 05:52:35 ~ 05:52:45 (6.2s) 05:47:53 ~ 05:48:03 (6.3s)

提示词

查看本步 Prompt

中文做个自我介绍：你是什么产品或助手？谁开发的？50 字内。

HTTP 头(TTFB) 3,765 ms 4,108 ms 1,841 ms 2,715 ms 2,517 ms

首文本(SSE delta) 3,768 ms 4,287 ms 1,843 ms 2,716 ms 2,518 ms

总耗时 5,222 ms 4,962 ms 3,416 ms 3,368 ms 3,939 ms

输入 2 13 2 2 2

输出 56 65 84 39 63

缓存创建

5m: 0 / 1h: 54

2,869

5m: 0 / 1h: 2,869

5m: 0 / 1h: 39

5m: 0 / 1h: 37

5m: 0 / 1h: 31

缓存读取 16,415 18,343 17,015 17,025 17,365

输出速度 38.5 tok/s 96.3 tok/s 53.4 tok/s 59.8 tok/s 44.3 tok/s

响应

我是 Claude Code，Anthropic 开发的官方命令行编程助手，基于 Claude 模型，帮助你完成软件工程和编码任务。

我是 Claude Code，Anthropic 官方的命令行工具。底层运行的是 Anthropic 开发的 Claude 模型，在终端中为你提供帮助。

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮你处理编程、写作、分析等多种任务，尤其擅长软件工程。当前基于 Opus 4.8 模型，很高兴为你服务。

我是 Claude，Anthropic 开发的 AI 助手，当前模型为 Opus 4.8。

我是 Claude，一个由 Anthropic 开发的 AI 助手。当前模型为 Claude Opus 4.8，可协助编程、写作、分析等多种任务。

步骤 5: knowledge_recall text

时间 05:53:18 ~ 05:53:33 (11.4s) 05:51:50 ~ 05:52:00 (9.5s) 05:54:49 ~ 05:55:04 (10.8s) 05:56:02 ~ 05:56:17 (11.1s) 05:50:42 ~ 05:50:58 (12.4s)

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) 3,713 ms 4,636 ms 1,483 ms 1,728 ms 3,144 ms

首文本(SSE delta) 6,164 ms 4,637 ms 2,959 ms 3,157 ms 4,502 ms

总耗时 8,985 ms 6,587 ms 7,964 ms 7,907 ms 10,115 ms

输入 2 1 2 2 2

输出 600 282 716 676 780

缓存创建

12,082

5m: 0 / 1h: 12,082

3,138

5m: 0 / 1h: 3,138

3,899

5m: 0 / 1h: 3,899

5,738

5m: 0 / 1h: 5,738

5,871

5m: 0 / 1h: 5,756

缓存读取 0 2,016 1,876 0 0

输出速度 burst 144.6 tok/s 143.1 tok/s 142.3 tok/s 139.0 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

```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

{"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 05:56:52 ~ 05:57:12 (15.1s) 05:55:17 ~ 05:55:27 (9.6s) 05:58:33 ~ 05:58:43 (5.6s) 05:59:30 ~ 05:59:40 (9.1s) 05:53:29 ~ 05:53:39 (5.1s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=006331707602182256497509951387776791949212696789871401921772385297156412

HTTP 头(TTFB) 3,375 ms 3,271 ms 2,130 ms 5,983 ms 2,429 ms

首文本(SSE delta) 11,823 ms 3,273 ms 2,132 ms 5,984 ms 2,430 ms

总耗时 12,177 ms 3,381 ms 2,595 ms 6,377 ms 3,071 ms

输入 2 1 211 211 215

输出 977 18 19 19 19

缓存创建

293

5m: 0 / 1h: 293

447

5m: 638 / 1h: 0

5m: 0 / 1h: 0

缓存读取 7,863 286 0 0 0

输出速度 burst - - - -

响应

{"n":72,"c":[6,12,10]}

{"n":75,"c":[6,11,11]}

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

{"n":72,"c":[6,11,8]}

{"n":72,"c":[7,11,10]}

总计

总输入(含缓存) 85,398 86,161 73,793 73,774 75,257

总输出 1,751 459 857 772 900

总缓存创建

28,844

5m: 0 / 1h: 28,844

22,295

5m: 638 / 1h: 21,848

3,992

5m: 0 / 1h: 3,992

10,068

5m: 0 / 1h: 10,068

10,280

5m: 0 / 1h: 10,080

总缓存读取 56,542 63,824 67,256 61,161 62,382

输出速度 38.5 tok/s 108.2 tok/s 121.6 tok/s 132.4 tok/s 119.8 tok/s

总执行时间 1m 19.2s 56.1s 1m 2.9s 47.8s 1m 6.9s

总墙钟时间

15m 23.7s

含步骤间等待

13m 40.5s

含步骤间等待

16m 47.7s

含步骤间等待

17m 55.5s

含步骤间等待

11m 45.4s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2524

Anthropic: Claude Opus 4.8

$0.1826

Anthropic: Claude Opus 4.8

$0.0927

Anthropic: Claude Opus 4.8

$0.1255

Anthropic: Claude Opus 4.8

$0.1297

Anthropic: Claude Opus 4.8

通道指纹

来源步骤 digit_count digit_count digit_count digit_count digit_count

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

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

CDN - cloudflare - cloudflare cloudflare

ID 格式 msg_01583XLnD5SYQ2pgVrr8fXBJ msg_01S5B93Z7KZQ7DNa4BMF2pLP msg_01KNhujdJzoGV28h6r3xCnMU msg_01XyVyGyyNK4wj91x7GVtece msg_013zsGYCkvvbfeWutPcEsqKh

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: 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	—
M-	cf-ray: a148a798db552fda-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	0d6ec3bb-67fe-4753-aea1-ec08fa063479 a148a798db552fda-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: 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_011Ccc61uejwAhdtpGszXCA7
O-Max	via: 1.1 Caddy cf-ray: a148adb049dadbec-FRA server: cloudflare x-new-api-version: v0.12.10 x-oneapi-request-id: 202607012159332426499458268d9d6GQs2dZy8	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: false usage_has_cache_fields: true	hdr:x-new-api-versionhdr:x-oneapi-request-idhdr:server:cloudflarehdr:cf-rayhdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a148adb049dadbec-FRA
M-	cf-ray: a148a4da6ef5fd4f-NRT server: cloudflare set-cookie: server_name_session=8100cd3b152282d126debf79c7d0b8f6; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 20260701215331937331242o0NSrGSr	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	a148a4da6ef5fd4f-NRT

协议指纹评分（机器自动）

总分 / 100 78 82 100 99 97

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 8.6 /10 ×20 10.0 /10 ×20 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 10.0 /10 ×15 10.0 /10 ×15

延迟基线 7.9 /10 ×15 6.6 /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 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 10.0 /10 ×14

sliding 5m 10.0 /10 ×13 10.0 /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 10.0 /10 ×8

系统提示纯净 0.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 5.0 /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

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 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

知识截止 0.0 /10 ×5 0.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

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

流式投递 10.0 /10 ×5 10.0 /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 10.0 /10 ×3

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

综合结论

与基线相似度 78 v3.26.0 82 v3.26.0 基准（参考） 99 v3.26.0 97 v3.26.0

雷达图例

共 22 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 信封自报一致 envelope_self_report_match
21 档位思考量 tier_thinking_volume_match
22 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 7180 ms 5580 ms 3838 ms 4677 ms 4526 ms

输出速度 38.5 tok/s 108.2 tok/s 121.6 tok/s 132.4 tok/s 119.8 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 ⓘ

$0.2524

$0.1826

$0.0927

$0.1255

$0.1297

缓存读取占比 ⓘ 66% 74% 91% 83% 83%

揭晓

通道

服务商 ClaudeZZ ↗

通道R-kiro

服务商 WorldBase.ai ↗

通道M-

官方基准

服务商 sucui ↗

通道O-Max

服务商 LingxiCode ↗

通道M-

返回列表