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。

4 通道 · 6 步骤

2026/06/28 运行日期

claude-opus-4-8 官方基准

1 通道分数 < 70 另含 2 任务失败

已揭晓揭晓进度

⚠ 1 通道不可测 · 已从主对比表移除点击展开原因

Aiberm / cc
```
评测失败
```

步骤 / 指标

官方基准

luminai

M-Kiro/Antigravity

SSSAiCode

O-Max-L2

ClaudeCN

O-Max

基本信息

服务名 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

客户端 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 05:11:52 ~ 05:11:57 (4.0s) 05:11:44 ~ 05:11:54 (5.1s) 05:11:50 ~ 05:11:55 (3.5s) 05:11:48 ~ 05:11:58 (6.6s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,858 ms 3,758 ms 2,132 ms 3,034 ms

首文本(SSE delta) 2,860 ms 3,759 ms 2,202 ms 3,040 ms

总耗时 2,996 ms 3,810 ms 2,280 ms 3,189 ms

输入 1,454 1,643 1,428 1,454

输出 4 1 4 4

缓存创建

30,222

5m: 0 / 1h: 30,222

1,888

5m: 1,888 / 1h: 0

29,763

5m: 0 / 1h: 29,763

30,225

5m: 0 / 1h: 30,225

缓存读取 0 16,997 0 0

响应

pong

pong

pong

pong

步骤 2: identity text

时间 05:15:50 ~ 05:15:55 (3.4s) 05:14:51 ~ 05:15:01 (5.6s) 05:15:23 ~ 05:15:28 (3.0s) 05:14:29 ~ 05:14:39 (5.0s)

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) 2,074 ms 4,409 ms 1,806 ms 3,455 ms

首文本(SSE delta) 2,075 ms 4,411 ms 2,144 ms 3,456 ms

总耗时 2,460 ms 4,463 ms 2,279 ms 3,954 ms

输入 2 1,657 2 2

输出 28 23 28 28

缓存创建

1,634

5m: 0 / 1h: 1,634

1,904

5m: 1,904 / 1h: 0

1,608

5m: 0 / 1h: 1,608

31,858

5m: 0 / 1h: 31,858

缓存读取 30,222 17,145 29,763 0

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

步骤 3: cutoff text

时间 05:18:47 ~ 05:18:52 (4.5s) 05:18:31 ~ 05:18:46 (12.1s) 05:16:35 ~ 05:16:40 (3.0s) 05:17:29 ~ 05:17:34 (4.5s)

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) 3,060 ms 10,823 ms 1,910 ms 2,969 ms

首文本(SSE delta) 3,062 ms 10,824 ms 2,170 ms 2,979 ms

总耗时 3,401 ms 10,883 ms 2,249 ms 3,211 ms

输入 2 1,662 2 2

输出 6 4 6 6

缓存创建

5m: 0 / 1h: 62

1,910

5m: 1,910 / 1h: 0

5m: 0 / 1h: 62

5m: 0 / 1h: 64

缓存读取 31,856 17,193 31,371 31,858

响应

2026-01

2026-01

2025-01

2026-01

步骤 4: identity_free text

时间 05:22:10 ~ 05:22:15 (4.2s) 05:21:27 ~ 05:21:37 (5.2s) 05:18:43 ~ 05:18:48 (3.3s) 05:21:25 ~ 05:21:35 (5.2s)

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) 2,257 ms 3,826 ms 2,018 ms 3,251 ms

首文本(SSE delta) 2,260 ms 4,050 ms 2,525 ms 3,275 ms

总耗时 3,213 ms 4,110 ms 2,549 ms 3,994 ms

输入 2 1,665 2 2

输出 58 28 44 47

缓存创建

5m: 0 / 1h: 35

1,914

5m: 1,914 / 1h: 0

5m: 0 / 1h: 31

5m: 0 / 1h: 37

缓存读取 31,918 17,229 31,433 31,922

输出速度 60.9 tok/s - burst 65.4 tok/s

响应

我是 Claude，由 Anthropic 打造的 AI 助手，当前模型为 Claude Opus 4.8，可协助编程、研究与各类任务。

我是 Claude Opus 4.8，属于 Claude 产品系列，由 Anthropic 开发。

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。

我叫 Claude，是 Anthropic 公司开发的 AI 助手，当前模型为 Claude Opus 4.8。

步骤 5: knowledge_recall text

时间 05:23:33 ~ 05:23:43 (9.0s) 05:23:57 ~ 05:24:12 (10.9s) 05:20:22 ~ 05:20:32 (6.5s) 05:24:30 ~ 05:24:40 (9.6s)

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) 1,535 ms 6,181 ms 2,972 ms 2,513 ms

首文本(SSE delta) 5,883 ms 6,182 ms 5,573 ms 5,804 ms

总耗时 7,970 ms 9,781 ms 5,719 ms 8,349 ms

输入 2 268 2 2

输出 684 393 320 672

缓存创建

4,765

5m: 0 / 1h: 4,765

308

5m: 308 / 1h: 0

4,262

5m: 0 / 1h: 4,262

4,760

5m: 0 / 1h: 4,760

缓存读取 0 2,772 0 0

输出速度 burst 109.2 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

{"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 05:24:56 ~ 05:25:01 (2.3s) 05:27:09 ~ 05:27:24 (14.0s) 05:23:45 ~ 05:23:50 (1.0s) 05:26:06 ~ 05:26:11 (3.1s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=956752307715794489832407601665950661012891457073736172469586776452688095

HTTP 头(TTFB) 1,032 ms 6,698 ms - 1,766 ms

首文本(SSE delta) 1,033 ms 6,699 ms - 1,767 ms

总耗时 1,372 ms 12,726 ms - 2,147 ms

输入 211 190 - 211

输出 19 102 - 19

缓存创建

5m: 0 / 1h: 0

缓存读取 0 - - 0

输出速度 - 16.9 tok/s - -

响应

{"n":72,"c":[6,9,4]}

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

失败

评测失败

{"n":72,"c":[6,11,5]}

总计

总输入(含缓存)

132,387

86,345

129,729

仅前 6 步

132,397

总输出

799

551

402

仅前 6 步

776

总缓存创建

36,718

5m: 0 / 1h: 36,718

7,924

5m: 7,924 / 1h: 0

35,726

5m: 0 / 1h: 35,726

仅前 6 步

66,944

5m: 0 / 1h: 66,944

总缓存读取

93,996

71,336

92,567

仅前 6 步

63,780

输出速度

60.9 tok/s

51.4 tok/s

burst

仅前 6 步

65.4 tok/s

总执行时间

27.4s

52.9s

20.3s

仅前 6 步

34.0s

总墙钟时间

13m 8.9s

含步骤间等待

15m 39.9s

含步骤间等待

11m 59.9s

含步骤间等待

仅前 6 步

14m 22.9s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.3048

Anthropic: Claude Opus 4.8

$0.1158

Anthropic: Claude Opus 4.8

$0.2868

Anthropic: Claude Opus 4.8 (部分)

$0.4781

Anthropic: Claude Opus 4.8

通道指纹

来源步骤 digit_count digit_count knowledge_recall digit_count

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

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

CDN - cloudflare - cloudflare

ID 格式 msg_019dR9SF6F3ufaq8ip222AB4 msg_01mWejdU9wDLYXq8gRtheumQmc msg_01Q6hkNqUW2iTNkR8zpX5Qv5 msg_01UJw15RRFShSvPFhJAMw7Hr

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_011CcUUDR5dp9AqE8so82HQA
M-Kiro/Antigravity	cf-ray: a12788c49c030b8c-AMS server: cloudflare x-oneapi-request-id: 202606272127109234580158268d9d6Cj8qAVvy	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	a12788c49c030b8c-AMS
O-Max-L2	cf-ray: a1277ed2cec1991f-LAX server: Photon-Edge set-cookie: _cfuvid=LC_NI.IyjZZ0WoutDeumSAqrO8V3IGEmOEMWIUFBfHo-1782595223.4910212-1.0.1.1-gHUYitrneE0SdZKgSnFsICew9XQ0Jn8LcBtUGiNNTSA; 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	1782595223396-mg308mxeuhac1 req_011CcUTsDTakJE1amKLeBvaK a1277ed2cec1991f-LAX
O-Max	cf-ray: a12787397a58f5f2-AMS server: cloudflare x-oneapi-request-id: 20260628052608151600378wMPJz6ta	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	a12787397a58f5f2-AMS

协议指纹评分（机器自动）

总分 / 100 100 73 91

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 — 6.0 /10 ×20

缓存 TTL 一致 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

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 — 7.7 /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 — 10.0 /10 ×8

身份(结构) 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

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 0.0 /10 ×5 — 10.0 /10 ×5

延迟基线 10.0 /10 ×5 8.1 /10 ×5 — 10.0 /10 ×5

流式投递 10.0 /10 ×5 0.0 /10 ×5 — 10.0 /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

综合结论

与基线相似度基准（参考） 73 v3.25.0 执行失败，不评分 91 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 3569 ms 7629 ms 3015 ms 4141 ms

输出速度 60.9 tok/s 51.4 tok/s burst 65.4 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

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.3048

$0.1158

$0.2868 (部分)

$0.4781

缓存读取占比 ⓘ 71% 83% 71% (部分) 48%

揭晓

通道

官方基准

服务商 luminai ↗

通道M-Kiro/Antigravity

服务商 SSSAiCode ↗

通道O-Max-L2

服务商 ClaudeCN ↗

通道O-Max

返回列表