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

2026/06/23 运行日期

claude-opus-4-8 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

ClaudeZZ

R-kiro

官方基准

SSSAiCode

O-Max

0-0

O-Max

OmniaKey

O-Max

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 v2.1.160 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 05:59:36 ~ 05:59:41 (4.7s) 05:59:38 ~ 05:59:43 (3.8s) 05:59:34 ~ 05:59:39 (4.5s) 05:59:30 ~ 05:59:40 (5.8s) 05:59:32 ~ 05:59:37 (4.8s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,338 ms 2,445 ms 2,137 ms 3,729 ms 3,246 ms

首文本(SSE delta) 3,340 ms 2,447 ms 2,287 ms 3,734 ms 3,248 ms

总耗时 3,726 ms 2,601 ms 2,670 ms 4,006 ms 3,431 ms

输入 2 1,454 1,428 2 1,454

输出 4 4 4 4 4

缓存创建

27,910

5m: 0 / 1h: 27,910

30,222

5m: 0 / 1h: 30,222

29,761

5m: 0 / 1h: 29,761

32,304

5m: 32,304 / 1h: 0

30,222

5m: 0 / 1h: 30,222

缓存读取 0 0 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 06:00:49 ~ 06:00:59 (5.4s) 06:02:08 ~ 06:02:13 (4.0s) 06:03:33 ~ 06:03:38 (3.9s) 06:00:51 ~ 06:01:01 (5.2s) 06:01:52 ~ 06:02:02 (5.4s)

提示词

查看本步 Prompt

输出 3 行 metadata 描述你自己（当前 API 模型）。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。禁止任何额外字符：
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) 4,114 ms 2,614 ms 2,562 ms 3,481 ms 3,978 ms

首文本(SSE delta) 4,116 ms 2,616 ms 3,004 ms 3,483 ms 3,979 ms

总耗时 4,424 ms 2,925 ms 3,080 ms 3,887 ms 4,360 ms

输入 2 2 2 2 2

输出 28 28 32 28 28

缓存创建

203

5m: 0 / 1h: 203

1,643

5m: 0 / 1h: 1,643

1,617

5m: 0 / 1h: 1,617

32,516

5m: 0 / 1h: 32,516

1,643

5m: 0 / 1h: 1,643

缓存读取 27,910 30,222 29,761 0 30,222

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-4-1-20250805

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 06:02:26 ~ 06:02:36 (5.0s) 06:03:30 ~ 06:03:35 (3.2s) 06:06:34 ~ 06:06:39 (3.6s) 06:02:28 ~ 06:02:33 (4.6s) 06:04:41 ~ 06:04:51 (6.2s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 3,678 ms 2,012 ms 2,559 ms 3,025 ms 4,753 ms

首文本(SSE delta) 3,679 ms 2,014 ms 2,667 ms 3,026 ms 4,754 ms

总耗时 4,120 ms 2,219 ms 2,834 ms 3,185 ms 5,002 ms

输入 2 2 2 2 2

输出 6 6 6 6 6

缓存创建

5m: 0 / 1h: 89

5m: 0 / 1h: 62

5m: 0 / 1h: 68

3,796

5m: 0 / 1h: 3,796

5m: 0 / 1h: 62

缓存读取 28,113 31,865 31,378 28,785 31,865

响应

2025-03

2026-01

2025-01

2026-01

2026-01

步骤 4: identity_free text

时间 06:04:26 ~ 06:04:36 (5.4s) 06:05:27 ~ 06:05:32 (4.1s) 06:10:07 ~ 06:10:12 (4.4s) 06:04:44 ~ 06:04:54 (5.7s) 06:06:53 ~ 06:07:03 (7.2s)

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) 3,538 ms 2,371 ms 2,533 ms 3,298 ms 5,203 ms

首文本(SSE delta) 3,539 ms 2,373 ms 3,611 ms 3,299 ms 5,204 ms

总耗时 4,405 ms 2,941 ms 3,670 ms 4,240 ms 6,107 ms

输入 2 2 2 2 2

输出 56 44 67 62 63

缓存创建

5m: 0 / 1h: 52

5m: 0 / 1h: 35

5m: 0 / 1h: 39

32,590

5m: 32,590 / 1h: 0

5m: 0 / 1h: 37

缓存读取 28,202 31,927 31,446 0 31,927

输出速度 64.7 tok/s 77.5 tok/s - 65.9 tok/s 69.8 tok/s

响应

我是 Claude，由 Anthropic 开发的 AI 助手，当前模型为 Claude Opus 4.8，属于 Claude 系列产品。

我是 Claude，由 Anthropic 打造的 AI 助手，当前模型为 Claude Opus 4.8。

我是 Claude，由 Anthropic 开发的人工智能助手。我能协助编程、写作、分析和回答各类问题，致力于提供有帮助、诚实且安全的服务。

我是 Claude，一个由 Anthropic 开发的人工智能助手，基于 Claude Opus 4.8 模型，可以帮助你完成编程和各类任务。

我是 Claude，具体型号为 Opus 4.8（API 标识 claude-opus-4-8），属于 Claude 产品系列，由 Anthropic 公司开发。

步骤 5: knowledge_recall text

时间 06:07:08 ~ 06:07:18 (8.6s) 06:09:23 ~ 06:09:33 (8.5s) 06:14:07 ~ 06:14:22 (13.0s) 06:06:55 ~ 06:07:10 (10.4s) 06:10:18 ~ 06:10:28 (9.5s)

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) 4,957 ms 2,381 ms 2,436 ms 2,881 ms 3,233 ms

首文本(SSE delta) 4,958 ms 5,183 ms 11,427 ms 4,647 ms 6,041 ms

总耗时 7,823 ms 7,388 ms 12,080 ms 8,943 ms 8,316 ms

输入 2 2 2 2 2

输出 348 625 863 700 689

缓存创建

11,113

5m: 0 / 1h: 11,113

4,760

5m: 0 / 1h: 4,760

4,306

5m: 0 / 1h: 4,306

5,357

5m: 5,357 / 1h: 0

4,781

5m: 0 / 1h: 4,781

缓存读取 0 0 0 0 0

输出速度 121.5 tok/s burst burst burst 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

{"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 06:09:56 ~ 06:10:11 (10.9s) 06:13:32 ~ 06:13:37 (2.9s) 06:16:08 ~ 06:16:13 (1.1s) 06:08:16 ~ 06:08:21 (3.2s) 06:12:31 ~ 06:12:36 (4.0s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=816100387399214141976422346890429840881114253513392254845220184383974062

HTTP 头(TTFB) 9,746 ms 1,468 ms - 1,687 ms 2,432 ms

首文本(SSE delta) 9,748 ms 1,473 ms - 1,688 ms 2,433 ms

总耗时 10,079 ms 1,879 ms - 2,031 ms 2,802 ms

输入 2 211 - 793 211

输出 774 19 - 19 19

缓存创建

5,088

5m: 0 / 1h: 5,088

5m: 0 / 1h: 0

缓存读取 3,067 0 - 0 0

输出速度 burst - - - -

响应

{"n":72,"c":[6,3,7]}

{"n":70,"c":[5,5,6]}

失败

评测失败

{"n":70,"c":[6,5,6]}

{"n":70,"c":[6,5,6]}

总计

总输入(含缓存)

131,759

132,409

129,812

仅前 6 步

136,151

132,432

总输出

1,216

726

972

仅前 6 步

819

809

总缓存创建

44,455

5m: 0 / 1h: 44,455

36,722

5m: 0 / 1h: 36,722

35,791

5m: 0 / 1h: 35,791

仅前 6 步

106,563

5m: 70,251 / 1h: 36,312

36,745

5m: 0 / 1h: 36,745

总缓存读取

87,292

94,014

92,585

仅前 6 步

28,785

94,014

输出速度

108.3 tok/s

77.5 tok/s

burst

仅前 6 步

65.9 tok/s

69.8 tok/s

总执行时间

40.0s

26.5s

30.6s

仅前 6 步

35.0s

37.0s

总墙钟时间

10m 35.0s

含步骤间等待

13m 58.9s

含步骤间等待

16m 39.0s

含步骤间等待

仅前 6 步

8m 50.8s

含步骤间等待

13m 3.9s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.3519

Anthropic: Claude Opus 4.8

$0.3030

Anthropic: Claude Opus 4.8

$0.3015

Anthropic: Claude Opus 4.8 (部分)

$0.5403

Anthropic: Claude Opus 4.8

$0.3053

Anthropic: Claude Opus 4.8

通道指纹

来源步骤 digit_count digit_count knowledge_recall digit_count digit_count

平台 direct (100%) 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%) Anthropic 直连 (100%)

CDN - - - - cloudflare

ID 格式 msg_01GTj5wpRCy3Q6WdYpectMkh msg_016TVjepoZKNKyFnPeCLm4q8 msg_011exuGjwNfxPyqvWCgodfAK msg_01P7C1FDFENVC8ZDt7hP89Ft msg_01R761qMWuY2CTwNX1DxYJ9v

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	via: 1.1 google cf-ray: 4ca9e4f45d8e29be0c7cc4cd6c344375-IAD server: nginx	id_format: msg_prefix via_chain: ["1.1 google"] extra_fields: ["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_4ca9e4f45d8e29be0c7cc4cd6c344375 4ca9e4f45d8e29be0c7cc4cd6c344375-IAD
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_011CcK4soP46S1FCaYSPUnvS
O-Max	cf-ray: a0fe9aaf89428626-LAX server: Photon-Edge set-cookie: _cfuvid=g9p6p18l6V1CF1JwH3N4NrADjOCc8atMXI7YBIUvhug-1782166448.5642173-1.0.1.1-DR6f1spg.efJuYyrIw5EWmegIuYM5bvKQYtbcbF__YM; HttpOnly; SameSite=None; Secure; Path=/; Domain=api.anthropic.com	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	1782166448524-35hm8vlrhfp92 req_011CcK4vNQTuP3V9jKeV9eGv a0fe9aaf89428626-LAX
O-Max	via: 1.1 Caddy server: uvicorn, nginx/1.24.0 (Ubuntu)	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:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	d7c1827d51ce56e694f7f99f9e24d0c3
O-Max	cf-ray: a0fe98597e78f5b4-AMS server: cloudflare x-oneapi-request-id: 202606222212333894470708268d9d6GR9vdKAP	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	a0fe98597e78f5b4-AMS

协议指纹评分（机器自动）

总分 / 100 83 100 75 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

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

缓存 TTL 一致 10.0 /10 ×15 10.0 /10 ×15 — 4.0 /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 10.0 /10 ×13 — 5.3 /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 0.0 /10 ×8 10.0 /10 ×8 — 5.0 /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 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

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 10.0 /10 ×5 10.0 /10 ×5 — 10.0 /10 ×5 9.3 /10 ×5

流式投递 10.0 /10 ×5 10.0 /10 ×5 — 6.7 /10 ×5 10.0 /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 83 v3.24.0 基准（参考）执行失败，不评分 75 v3.24.0 97 v3.24.0

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 缓存连续 cache_continuity_intra
14 知识召回 knowledge_recall_match
15 延迟基线 latency_baseline_match
16 原生 msg-ID anthropic_msg_id_format
17 Sliding 缓存 cache_sliding_correctness
18 结构化身份 identity_structured_match
19 信封自报一致 envelope_self_report_match
20 中文遵循 instruction_following_lang
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 5763 ms 3326 ms 4867 ms 4382 ms 5003 ms

输出速度 108.3 tok/s 77.5 tok/s burst 65.9 tok/s 69.8 tok/s

完成步骤 6/6 成功 6/6 成功 5/6 成功 · 1 失败 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 ⓘ

$0.3519

$0.3030

$0.3015 (部分)

$0.5403

$0.3053

缓存读取占比 ⓘ 66% 71% 71% (部分) 21% 71%

揭晓

通道

服务商 ClaudeZZ ↗

通道R-kiro

官方基准

服务商 SSSAiCode ↗

通道O-Max

服务商 0-0 ↗

通道O-Max

服务商 OmniaKey ↗

通道O-Max

返回列表