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-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

SAIAi

O-Max

SSSAiCode

R-Kiro

FoxCode

R-ultra

SSSAiCode

O-Max

官方基准

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 11:01:30 ~ 11:01:35 (3.7s) 11:01:32 ~ 11:01:37 (3.2s) 11:01:26 ~ 11:01:31 (4.7s) 11:01:28 ~ 11:01:33 (3.4s) 11:01:34 ~ 11:01:39 (3.2s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,131 ms 2,241 ms 2,246 ms 2,024 ms 1,924 ms

首文本(SSE delta) 2,133 ms 2,281 ms 2,388 ms 2,084 ms 1,927 ms

总耗时 2,401 ms 2,288 ms 2,611 ms 2,169 ms 2,140 ms

输入 1,454 0 23 1,428 1,454

输出 4 4 2 4 4

缓存创建

30,222

5m: 0 / 1h: 30,222

23,537

5m: 23,537 / 1h: 0

29,763

5m: 0 / 1h: 29,763

5m: 0 / 1h: 0

缓存读取 0 0 0 0 30,222

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 11:05:31 ~ 11:05:36 (4.2s) 11:02:45 ~ 11:02:55 (6.6s) 11:03:54 ~ 11:03:59 (3.9s) 11:03:33 ~ 11:03:38 (2.8s) 11:03:41 ~ 11:03:46 (3.9s)

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。请按下列 3 行格式精确填空作答（label 不变，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) 2,775 ms 5,611 ms 1,871 ms 1,579 ms 2,527 ms

首文本(SSE delta) 2,778 ms 5,638 ms 1,872 ms 1,904 ms 2,529 ms

总耗时 3,169 ms 5,841 ms 3,093 ms 2,048 ms 2,911 ms

输入 2 0 70 2 2

输出 28 28 64 28 28

缓存创建

5m: 0 / 1h: 0

148

5m: 148 / 1h: 0

1,608

5m: 0 / 1h: 1,608

1,634

5m: 0 / 1h: 1,634

缓存读取 31,856 0 23,537 29,763 30,222

输出速度 - - 52.4 tok/s - -

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

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 11:08:37 ~ 11:08:42 (3.0s) 11:05:49 ~ 11:05:54 (3.2s) 11:07:07 ~ 11:07:12 (4.3s) 11:05:53 ~ 11:06:03 (4.9s) 11:07:33 ~ 11:07:38 (3.6s)

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) 1,807 ms 2,220 ms 2,306 ms 3,834 ms 2,003 ms

首文本(SSE delta) 1,809 ms 2,354 ms 2,307 ms 4,001 ms 2,005 ms

总耗时 1,996 ms 2,358 ms 3,324 ms 4,150 ms 2,598 ms

输入 2 0 24 2 2

输出 6 8 36 6 6

缓存创建

5m: 0 / 1h: 62

24,110

5m: 24,110 / 1h: 0

5m: 0 / 1h: 64

5m: 0 / 1h: 54

缓存读取 31,856 0 0 31,371 31,856

输出速度 - - 35.4 tok/s - -

响应

2026-01

2026-01

2026-01

2025-03

2026-01

步骤 4: identity_free text

时间 11:10:11 ~ 11:10:16 (4.4s) 11:07:21 ~ 11:07:31 (8.6s) 11:09:48 ~ 11:09:58 (5.6s) 11:08:25 ~ 11:08:30 (3.7s) 11:09:27 ~ 11:09:37 (5.4s)

提示词

查看本步 Prompt

用中文回答：你是谁？由谁开发？50 字以内。

HTTP 头(TTFB) 2,448 ms 7,306 ms 2,901 ms 2,033 ms 3,636 ms

首文本(SSE delta) 2,450 ms 7,308 ms 2,902 ms 2,537 ms 3,637 ms

总耗时 3,395 ms 7,754 ms 4,732 ms 2,603 ms 4,392 ms

输入 2 0 71 2 2

输出 51 44 80 44 51

缓存创建

5m: 0 / 1h: 31

335

5m: 335 / 1h: 0

31,463

5m: 0 / 1h: 31,463

5m: 0 / 1h: 37

缓存读取 31,918 0 24,110 0 31,910

输出速度 54.0 tok/s 98.7 tok/s 43.7 tok/s - 67.5 tok/s

响应

我是 Claude，由 Anthropic 开发的人工智能助手，可以帮你完成写作、编程、分析和问答等各类任务。

我是 Claude，由 Anthropic 开发的 AI 助手。我可以帮你处理编程、写作、分析、研究等各类任务。

我叫 Kiro，是由 Anthropic 开发的 Claude 模型（claude-opus-4-8）驱动的 AI 助手，属于 Claude 4产品系列。

我是 Claude，一个由 Anthropic 开发的人工智能助手，旨在帮助用户完成各类任务。

我是 Claude，属于 Claude 产品系列，由 Anthropic 公司开发。具体模型是 Claude Opus 4.8。

步骤 5: knowledge_recall text

时间 11:11:57 ~ 11:12:07 (8.0s) 11:10:31 ~ 11:10:46 (11.0s) 11:13:03 ~ 11:13:23 (19.7s) 11:09:38 ~ 11:09:43 (4.6s) 11:11:59 ~ 11:12:09 (9.1s)

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) 2,186 ms 4,679 ms 3,286 ms 1,166 ms 2,695 ms

首文本(SSE delta) 4,540 ms 5,383 ms 3,287 ms 3,788 ms 5,509 ms

总耗时 6,952 ms 10,210 ms 18,834 ms 3,856 ms 7,924 ms

输入 2 0 21 2 2

输出 655 342 1,474 320 654

缓存创建

4,754

5m: 0 / 1h: 4,754

4,736

5m: 4,736 / 1h: 0

4,261

5m: 0 / 1h: 4,261

2,980

5m: 0 / 1h: 2,980

缓存读取 0 0 0 0 1,772

输出速度 burst 70.9 tok/s 94.8 tok/s 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 11:14:52 ~ 11:14:57 (3.5s) 11:14:03 ~ 11:14:18 (11.3s) 11:17:01 ~ 11:18:01 (57.2s) 11:11:39 ~ 11:11:44 (0.9s) 11:15:52 ~ 11:15:57 (2.7s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=100175398399743121504364707124737708873544305428877571435341510381333816

HTTP 头(TTFB) 2,164 ms 4,675 ms 36,259 ms - 1,353 ms

首文本(SSE delta) 2,166 ms 10,356 ms 36,260 ms - 1,355 ms

总耗时 2,482 ms 10,499 ms 56,366 ms - 1,754 ms

输入 211 0 56 - 211

输出 19 20 2,675 - 19

缓存创建

5m: 0 / 1h: 0

153

5m: 153 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 0 - 0

输出速度 - - 133.0 tok/s - -

响应

{"n":70,"c":[8,8,2]}

{"n":72,"c":[7,11,3]}

Counting s = `100175398399743121504364707124737708873544305428877571435341510381333816`

- n = 72
- 0s at indices: 1,2,19,25,34,43,62 → **7**
- 7s at indices: 4,12,24,26,30,32,33,37,49,50,52 → **11**

失败

评测失败

{"n":70,"c":[5,11,3]}

总计

总输入(含缓存)

132,372

100,931

129,729

仅前 6 步

132,360

总输出

763

446

4,331

402

仅前 6 步

762

总缓存创建

35,069

5m: 0 / 1h: 35,069

53,019

5m: 53,019 / 1h: 0

67,159

5m: 0 / 1h: 67,159

仅前 6 步

4,705

5m: 0 / 1h: 4,705

总缓存读取

95,630

47,647

61,134

仅前 6 步

125,982

输出速度

54.0 tok/s

73.2 tok/s

109.0 tok/s

burst

仅前 6 步

67.5 tok/s

总执行时间

26.7s

43.8s

1m 35.5s

20.3s

仅前 6 步

28.0s

总墙钟时间

13m 27.0s

含步骤间等待

12m 46.0s

含步骤间等待

16m 35.2s

含步骤间等待

10m 16.0s

含步骤间等待

仅前 6 步

14m 23.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2944

Anthropic: Claude Opus 4.8

>=$0.0112

Anthropic: Claude Opus 4.8 (成本不完整)

$0.3405

Anthropic: Claude Opus 4.8

$0.4675

Anthropic: Claude Opus 4.8 (部分)

$0.1198

Anthropic: Claude Opus 4.8

通道指纹

来源步骤 digit_count digit_count digit_count knowledge_recall digit_count

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

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

CDN - - cloudflare - -

ID 格式 msg_01NoXssEsW3C8EHoSSsXLW8n msg_cd14f92f89cd41bf899f744b262087b6 msg_01JJzKLO7EV7jOAMHyWmElhw msg_01GAoD8SbHALDFTBsNfTh9Gm msg_01UoSeBqsnQTSkGsDdZLnVEe

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
O-Max	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_011CcMMfUKyoqADi4JLQQxsv
R-Kiro	server: Photon-Edge x-amzn-requestid: 71f78902-a3af-4757-ba2e-4758461d6446	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	1782270844467-775tbkcv9xtht 71f78902-a3af-4757-ba2e-4758461d6446
R-ultra	cf-ray: a10893c23a857343-NRT server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606240317026144991158268d9d6YwDJ1D7S	id_format: msg_prefix 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-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsupstream:anthropic-direct-from-body	a10893c23a857343-NRT
O-Max	cf-ray: a10888efba5c090c-LAX server: Photon-Edge set-cookie: _cfuvid=CF1LiFr4T88KOXf.oGa4waO47RkUXiL11JM2mepWP8s-1782270579.15609-1.0.1.1-RMqyV2TtaBsINHLc3IuPeQGIqgI0VLdcfVr04I3bU_8; 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	1782270579112-1ahh4h7jehdqw req_011CcMMGHQWsBteKagkotpt5 a10888efba5c090c-LAX
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_011CcMMjt7KyJmsXPufRMkdK

协议指纹评分（机器自动）

总分 / 100 100 55 69 100

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 5.5 /10 ×20 — 10.0 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 0.0 /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

缓存连续性 10.0 /10 ×14 5.0 /10 ×14 10.0 /10 ×14 — 10.0 /10 ×14

sliding 5m 10.0 /10 ×13 5.0 /10 ×13 7.7 /10 ×13 — 10.0 /10 ×13

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

系统提示纯净 10.0 /10 ×8 未计入 0.0 /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 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 0.0 /10 ×7 — 10.0 /10 ×7

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

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

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

Req-ID 透传 10.0 /10 ×4 0.0 /10 ×4 0.0 /10 ×4 — 10.0 /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 100 v3.24.0 55 v3.24.0 69 v3.24.0 执行失败，不评分基准（参考）

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 3399 ms 6492 ms 14827 ms 2965 ms 3620 ms

输出速度 54.0 tok/s 73.2 tok/s 109.0 tok/s burst 67.5 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 ⓘ

$0.2944

>=$0.0112 (成本不完整)

$0.3405

$0.4675 (部分)

$0.1198

缓存读取占比 ⓘ 72% - 47% 47% (部分) 95%

揭晓

通道

服务商 SAIAi ↗

通道O-Max

服务商 SSSAiCode ↗

通道R-Kiro

服务商 FoxCode ↗

通道R-ultra

服务商 SSSAiCode ↗

通道O-Max

官方基准

返回列表