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

2026/06/30 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

官方基准

AiTokensFlux

O-Pro

luminai

M-Kiro/Antigravity

ReClaude

O-Max-CP4

WorldBase.ai

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 v2.1.160 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 10:34:44 ~ 10:34:49 (3.2s) 10:34:42 ~ 10:34:47 (4.6s) 10:34:36 ~ 10:34:46 (5.1s) 10:34:38 ~ 10:34:43 (4.7s) 10:34:40 ~ 10:34:50 (7.8s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 1,939 ms 3,451 ms 3,637 ms 3,619 ms 4,942 ms

首文本(SSE delta) 1,942 ms 3,466 ms 3,639 ms 3,622 ms 4,943 ms

总耗时 2,118 ms 3,698 ms 3,686 ms 3,627 ms 5,878 ms

输入 1,454 1,454 1,643 1,454 1,048

输出 4 4 1 4 1

缓存创建

30,222

5m: 0 / 1h: 30,222

30,228

5m: 0 / 1h: 30,228

2,266

5m: 2,266 / 1h: 0

30,222

5m: 0 / 1h: 30,222

20,322

5m: 0 / 1h: 20,322

缓存读取 0 0 16,619 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 10:37:54 ~ 10:38:04 (5.9s) 10:36:25 ~ 10:36:30 (4.0s) 10:37:04 ~ 10:37:14 (5.2s) 10:37:49 ~ 10:37:59 (5.7s) 10:37:17 ~ 10:37:27 (5.8s)

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。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) 4,366 ms 2,530 ms 3,934 ms 4,336 ms 4,587 ms

首文本(SSE delta) 4,516 ms 2,570 ms 3,934 ms 4,342 ms 4,588 ms

总耗时 4,779 ms 3,206 ms 4,056 ms 4,704 ms 4,683 ms

输入 2 2 1,656 2 16

输出 28 28 23 28 14

缓存创建

1,630

5m: 0 / 1h: 1,630

1,630

5m: 0 / 1h: 1,630

2,285

5m: 2,285 / 1h: 0

1,630

5m: 0 / 1h: 1,630

1,159

5m: 0 / 1h: 1,159

缓存读取 30,222 30,228 16,757 30,222 20,322

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 10:40:51 ~ 10:40:56 (4.6s) 10:40:07 ~ 10:40:12 (4.8s) 10:38:25 ~ 10:38:35 (5.7s) 10:41:42 ~ 10:41:52 (5.4s) 10:39:09 ~ 10:39:19 (7.3s)

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) 3,410 ms 3,961 ms 3,823 ms 4,187 ms 6,306 ms

首文本(SSE delta) 3,411 ms 3,962 ms 3,824 ms 4,286 ms 6,307 ms

总耗时 3,629 ms 4,083 ms 3,824 ms 4,380 ms 6,345 ms

输入 2 2 1,661 2 24

输出 6 6 4 6 2

缓存创建

5m: 0 / 1h: 64

2,291

5m: 2,291 / 1h: 0

5m: 0 / 1h: 62

5m: 0 / 1h: 26

缓存读取 31,852 31,858 16,805 31,852 21,481

响应

2026-01

2026-01

2026-01

2026-01

2026-01

步骤 4: identity_free text

时间 10:44:40 ~ 10:44:45 (4.8s) 10:43:16 ~ 10:43:26 (5.8s) 10:39:42 ~ 10:39:52 (6.1s) 10:43:35 ~ 10:43:45 (7.0s) 10:41:29 ~ 10:41:39 (8.8s)

提示词

查看本步 Prompt

用中文说说你的身份：什么产品？由谁打造？50 字以内。

HTTP 头(TTFB) 2,733 ms 3,998 ms 4,175 ms 5,213 ms 7,326 ms

首文本(SSE delta) 2,735 ms 4,045 ms 4,705 ms 5,214 ms 7,347 ms

总耗时 3,863 ms 4,997 ms 4,771 ms 6,027 ms 7,486 ms

输入 2 2 1,665 2 32

输出 61 65 58 55 17

缓存创建

5m: 0 / 1h: 35

31,961

5m: 0 / 1h: 31,961

2,296

5m: 2,296 / 1h: 0

5m: 0 / 1h: 35

5m: 0 / 1h: 21

缓存读取 31,916 0 16,842 31,914 21,507

输出速度 54.1 tok/s 68.3 tok/s - 67.7 tok/s -

响应

我是 Claude，一个由 Anthropic 打造的 AI 助手，当前模型为 Claude Opus 4.8，可协助编程与各类软件工程任务。

我是 Claude，由 Anthropic 开发的人工智能助手，擅长对话、写作、编程和分析等任务，致力于提供有帮助、诚实且安全的回应。

我是 Claude，由 Anthropic 开发的 AI 助手。我的版本是 Claude Opus 4.8，可以帮你写代码、分析问题、做研究和处理各类任务。

我是 Claude，由 Anthropic 公司打造的人工智能助手，具体为 Opus 4.8 模型（claude-opus-4-8）。

我是 Claude，属于 Claude 产品系列，由 Anthropic 开发。

步骤 5: knowledge_recall text

时间 10:47:48 ~ 10:48:03 (10.1s) 10:46:25 ~ 10:46:35 (8.4s) 10:41:04 ~ 10:41:19 (12.4s) 10:47:02 ~ 10:47:12 (8.6s) 10:43:58 ~ 10:44:08 (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) 3,133 ms 1,459 ms 7,668 ms 3,557 ms 5,787 ms

首文本(SSE delta) 6,906 ms 3,419 ms 7,669 ms 4,727 ms 5,881 ms

总耗时 9,092 ms 7,466 ms 11,359 ms 7,602 ms 8,219 ms

输入 2 2 270 2 40

输出 693 694 403 502 319

缓存创建

4,766

5m: 0 / 1h: 4,766

4,774

5m: 0 / 1h: 4,774

372

5m: 372 / 1h: 0

4,758

5m: 0 / 1h: 4,758

3,253

5m: 0 / 1h: 3,253

缓存读取 0 0 2,728 0 0

输出速度 burst burst 109.2 tok/s burst 136.4 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 10:51:18 ~ 10:51:23 (3.2s) 10:50:30 ~ 10:50:40 (7.0s) 10:42:45 ~ 10:43:06 (17.1s) 10:51:01 ~ 10:53:57 (170.7s) 10:46:35 ~ 10:47:00 (23.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=434651870517005091876637031854037664620437190654269920743435592067451000

HTTP 头(TTFB) 1,690 ms 5,967 ms 5,861 ms - 2,708 ms

首文本(SSE delta) 1,692 ms 5,968 ms 5,862 ms - 2,806 ms

总耗时 2,211 ms 6,289 ms 15,969 ms - 22,687 ms

输入 211 211 189 - 1

输出 19 19 97 - 1,699

缓存创建

5m: 0 / 1h: 0

154

缓存读取 0 0 - - 97

输出速度 - - 9.6 tok/s - 85.5 tok/s

响应

{"n":72,"c":[10,8,4]}

{"n":72,"c":[10,9,4]}

{"n":72,"c":[13,8,5]}

失败

评测失败

Breaking s into groups of 10 for accurate counting:

```
4346518705  (1–10)
1700509187  (11–20)
6637031854  (21–30)
0376646204  (31–40)
3719065426  (41–50)
9920743435  (51–60)
5920674510  (61–70)
00

总计

总输入(含缓存)

132,380

132,416

86,345

132,157

仅前 6 步

89,503

总输出

811

816

586

595

仅前 6 步

2,052

总缓存创建

36,717

5m: 0 / 1h: 36,717

68,657

5m: 0 / 1h: 68,657

9,510

5m: 9,510 / 1h: 0

36,707

5m: 0 / 1h: 36,707

仅前 6 步

24,935

5m: 0 / 1h: 24,781

总缓存读取

93,990

62,086

69,751

93,988

仅前 6 步

63,407

输出速度

54.1 tok/s

68.3 tok/s

36.2 tok/s

67.7 tok/s

仅前 6 步

90.8 tok/s

总执行时间

31.8s

34.6s

51.6s

3m 22.1s

仅前 6 步

1m 3.1s

总墙钟时间

16m 40.0s

含步骤间等待

15m 58.9s

含步骤间等待

8m 30.0s

含步骤间等待

19m 19.3s

含步骤间等待

仅前 6 步

12m 20.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.3051

Anthropic: Claude Opus 4.8

$0.4889

Anthropic: Claude Opus 4.8

$0.1221

Anthropic: Claude Opus 4.8

$0.2986

Anthropic: Claude Opus 4.8 (部分)

$0.2437

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

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

CDN - - cloudflare - cloudflare

ID 格式 msg_016wLUEKvS12w7s8Q7k7Tff9 msg_019n6Ffv1zuzJUgBKPPMKCc3 msg_01sFcYC5LqJ1j3aiUeESMJAoPC msg_01SRLpHUVpB3SXed319tVE7b msg_62278b1c29b0411cae9a6ad4a028ceb2

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_011CcYgimWvbs7rD3hSkoffT
O-Pro	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-Kiro/Antigravity	cf-ray: a139d1d4ba2fe560-AMS server: cloudflare x-oneapi-request-id: 202606300242476360208558268d9d6koAl0sS2	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	a139d1d4ba2fe560-AMS
O-Max-CP4	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_011CcYgPz4hhg5x9YBC3RwLz
M-	cf-ray: a139d76d1d22fe9b-AMS server: cloudflare	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	hdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	ddc9364c-6d39-487d-80e9-c2fb6b863d42 a139d76d1d22fe9b-AMS

协议指纹评分（机器自动）

总分 / 100 100 88 74 73

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 5.3 /10 ×20 10.0 /10 ×20 — 10.0 /10 ×20

缓存 TTL 一致 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

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 7.7 /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 — 0.0 /10 ×8

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

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

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

延迟基线 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 — 6.5 /10 ×5

流式投递 10.0 /10 ×5 2.6 /10 ×5 0.0 /10 ×5 — 0.0 /10 ×5

Req-ID 透传 10.0 /10 ×4 0.0 /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

与基线相似度基准（参考） 88 v3.25.0 74 v3.25.0 执行失败，不评分 73 v3.25.0

雷达图例

共 21 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 档位思考量 tier_thinking_volume_match
21 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 4282 ms 4957 ms 7278 ms 5268 ms 9216 ms

输出速度 54.1 tok/s 68.3 tok/s 36.2 tok/s 67.7 tok/s 90.8 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 ⓘ

$0.3051

$0.4889

$0.1221

$0.2986 (部分)

$0.2437

缓存读取占比 ⓘ 71% 47% 81% 71% (部分) 71%

揭晓

通道

官方基准

服务商 AiTokensFlux ↗

通道O-Pro

服务商 luminai ↗

通道M-Kiro/Antigravity

服务商 ReClaude ↗

通道O-Max-CP4

服务商 WorldBase.ai ↗

通道M-

返回列表