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

2026/06/19 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

WorldBase.ai

M-

官方基准

luminai

M-Kiro/Antigravity

FastCode

ClaudeCN

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 15:04:51 ~ 15:05:06 (11.9s) 15:04:57 ~ 15:05:02 (3.5s) 15:04:53 ~ 15:05:03 (6.1s) 15:04:49 ~ 15:05:04 (12.0s) 15:04:55 ~ 15:05:05 (5.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 9,485 ms 1,720 ms 4,453 ms 9,570 ms 3,589 ms

首文本(SSE delta) 9,486 ms 1,721 ms 4,721 ms 9,879 ms 3,593 ms

总耗时 9,512 ms 1,853 ms 4,721 ms 10,594 ms 3,794 ms

输入 1,212 1,454 1,027 1,454 1,454

输出 2 4 1 4 4

缓存创建

23,025

5m: 23,025 / 1h: 0

30,222

5m: 0 / 1h: 30,222

2,925

5m: 2,925 / 1h: 0

30,222

5m: 0 / 1h: 30,222

30,223

5m: 0 / 1h: 30,223

缓存读取 0 0 16,576 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 15:09:01 ~ 15:09:11 (5.8s) 15:07:13 ~ 15:07:18 (3.5s) 15:06:27 ~ 15:06:42 (14.1s) 15:07:07 ~ 15:07:17 (6.3s) 15:08:21 ~ 15:08:26 (4.3s)

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) 3,719 ms 2,045 ms 12,709 ms 3,928 ms 2,810 ms

首文本(SSE delta) 3,720 ms 2,047 ms 12,711 ms 3,930 ms 2,811 ms

总耗时 4,259 ms 2,543 ms 12,813 ms 4,907 ms 3,313 ms

输入 1,701 2 1,036 2 2

输出 15 28 23 28 28

缓存创建

32,300

5m: 32,300 / 1h: 0

1,634

5m: 0 / 1h: 1,634

2,950

5m: 2,950 / 1h: 0

1,634

5m: 0 / 1h: 1,634

1,634

5m: 0 / 1h: 1,634

缓存读取 0 30,222 16,719 30,222 30,223

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 15:10:23 ~ 15:10:33 (5.2s) 15:10:53 ~ 15:10:58 (3.0s) 15:10:07 ~ 15:10:23 (10.9s) 15:10:09 ~ 15:10:19 (5.9s) 15:10:26 ~ 15:10:41 (14.0s)

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) 3,448 ms 1,702 ms 9,325 ms 3,216 ms 3,340 ms

首文本(SSE delta) 3,449 ms 1,704 ms 9,326 ms 3,217 ms 3,341 ms

总耗时 3,564 ms 1,917 ms 9,399 ms 3,457 ms 12,635 ms

输入 1,226 2 1,039 2 2

输出 3 6 4 6 6

缓存创建

540

5m: 540 / 1h: 0

5m: 0 / 1h: 62

2,958

5m: 2,958 / 1h: 0

5m: 0 / 1h: 64

5m: 0 / 1h: 62

缓存读取 22,736 31,856 16,768 31,856 31,857

响应

2026-01

2026-01

2026-01

2026-01

2026-01

步骤 4: identity_free text

时间 15:13:31 ~ 15:13:41 (6.1s) 15:13:17 ~ 15:13:27 (5.0s) 15:12:53 ~ 15:13:03 (8.0s) 15:12:51 ~ 15:13:01 (6.3s) 15:13:49 ~ 15:13:59 (6.5s)

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) 3,729 ms 3,212 ms 5,429 ms 4,355 ms 4,379 ms

首文本(SSE delta) 5,006 ms 3,214 ms 6,393 ms 5,148 ms 4,381 ms

总耗时 5,007 ms 4,025 ms 6,577 ms 5,149 ms 5,524 ms

输入 1,238 2 1,041 2 2

输出 45 59 47 46 67

缓存创建

23,517

5m: 23,517 / 1h: 0

5m: 0 / 1h: 35

2,964

5m: 2,964 / 1h: 0

5m: 0 / 1h: 31

5m: 0 / 1h: 35

缓存读取 0 31,918 16,800 31,920 31,919

输出速度 burst 72.7 tok/s - burst 58.6 tok/s

响应

我是 Claude，一个由 Anthropic 开发的 AI 助手。当前模型是 Claude Opus 4.8，可以帮你处理编程、写作、分析等各类任务。

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、写作与问答等任务。

我是 Claude Opus 4.8，由 Anthropic 打造的 AI 助手，可帮你处理编程、写作、分析等各类任务。

我是 Claude，一个由 Anthropic 开发的人工智能助手，基于 Claude Opus 4.8 模型。

我是 Claude，由 Anthropic 打造的 AI 助手，当前模型为 Claude Opus 4.8，可通过 Claude Code 等方式协助编程与各类任务。

步骤 5: knowledge_recall text

时间 15:17:01 ~ 15:17:21 (15.1s) 15:16:54 ~ 15:17:04 (8.6s) 15:14:42 ~ 15:15:07 (22.8s) 15:15:27 ~ 15:15:42 (10.5s) 15:17:08 ~ 15:17:23 (10.3s)

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) 8,023 ms 2,637 ms 18,563 ms 3,052 ms 4,862 ms

首文本(SSE delta) 8,024 ms 5,455 ms 18,764 ms 6,150 ms 6,929 ms

总耗时 13,832 ms 7,627 ms 20,712 ms 9,220 ms 9,191 ms

输入 195 2 169 2 2

输出 366 694 419 656 677

缓存创建

3,686

5m: 3,686 / 1h: 0

4,766

5m: 0 / 1h: 4,766

478

5m: 478 / 1h: 0

4,751

5m: 0 / 1h: 4,751

4,775

5m: 0 / 1h: 4,775

缓存读取 0 0 2,714 0 0

输出速度 63.0 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 15:20:16 ~ 15:20:36 (18.4s) 15:20:30 ~ 15:20:35 (2.5s) 15:18:23 ~ 15:18:38 (12.2s) 15:17:21 ~ 15:17:26 (4.5s) 15:19:03 ~ 15:19:08 (4.0s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=260148448119148676123134487877112774257134981736320818045976119784839969

HTTP 头(TTFB) 10,892 ms 1,057 ms 4,088 ms 2,157 ms 2,374 ms

首文本(SSE delta) 10,893 ms 1,059 ms 4,089 ms 2,159 ms 2,376 ms

总耗时 16,312 ms 1,385 ms 10,972 ms 3,342 ms 2,811 ms

输入 473 211 190 211 211

输出 42 19 74 19 19

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 - 0 0

输出速度 7.8 tok/s - 10.8 tok/s - -

响应

{"n":72,"c":[3,10,7]}

{"n":71,"c":[3,11,8]}

{"n":72,"c":[3,10,7]}

{"n":72,"c":[2,9,6]}

{"n":72,"c":[5,9,6]}

总计

总输入(含缓存) 111,849 132,388 86,354 132,373 132,401

总输出 473 810 568 759 801

总缓存创建

83,068

5m: 83,068 / 1h: 0

36,719

5m: 0 / 1h: 36,719

12,275

5m: 12,275 / 1h: 0

36,702

5m: 0 / 1h: 36,702

36,729

5m: 0 / 1h: 36,729

总缓存读取 22,736 93,996 69,577 93,998 93,999

输出速度 36.3 tok/s 72.7 tok/s 10.8 tok/s burst 58.6 tok/s

总执行时间 1m 2.6s 26.2s 1m 14.2s 45.6s 44.4s

总墙钟时间

15m 45.0s

含步骤间等待

15m 38.1s

含步骤间等待

13m 45.0s

含步骤间等待

12m 37.0s

含步骤间等待

14m 13.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.3779

Anthropic: Claude Opus 4.8

$0.3051

Anthropic: Claude Opus 4.8

$0.1194

Anthropic: Claude Opus 4.8

$0.3037

Anthropic: Claude Opus 4.8

$0.3049

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%) 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%) Anthropic 直连 (100%)

CDN cloudflare - cloudflare - cloudflare

ID 格式 msg_76e8c96409df43e99d9446cc2dd5cf66 msg_017QMTnzT7J821HWxfTxyhHN msg_01VLbFZFmidKM1ZdRAyrKPNfTo msg_012gRvAFUZEpaenwi7BbTepC msg_01XXjs7pmMEx2jv2JQRDScCf

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
M-	cf-ray: a0e0c53cbaa80b48-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	6d2ffded-5b7b-4f40-8b28-a12e53c68f14 a0e0c53cbaa80b48-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_011CcCDMJRxKCf1Yv8RWzg1E
M-Kiro/Antigravity	cf-ray: a0e0c2753ac7b8ed-AMS server: cloudflare x-oneapi-request-id: 202606190718249340366238268d9d6F7pt3SWS	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	a0e0c2753ac7b8ed-AMS
cc	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	—
O-Max	cf-ray: a0e0c36e5831b936-AMS server: cloudflare x-oneapi-request-id: 20260619151905278848644ff8ukzNS	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	a0e0c36e5831b936-AMS

协议指纹评分（机器自动）

总分 / 100 59 100 72 98 98

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 0.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20

缓存 TTL 一致 0.0 /10 ×15 10.0 /10 ×15 0.0 /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 6.7 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

sliding 5m 5.3 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13

原生 msg-ID 0.0 /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 0.0 /10 ×8 10.0 /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 10.0 /10 ×7

知识截止 10.0 /10 ×7 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 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

inference_geo 10.0 /10 ×5 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 2.3 /10 ×5 9.7 /10 ×5 9.9 /10 ×5

流式投递 0.0 /10 ×5 10.0 /10 ×5 0.0 /10 ×5 10.0 /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 0.0 /10 ×4

stop_reason 10.0 /10 ×3 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 10.0 /10 ×2

综合结论

与基线相似度 59 v3.24.0 基准（参考） 72 v3.24.0 98 v3.24.0 98 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 8748 ms 3225 ms 10866 ms 6112 ms 6211 ms

输出速度 36.3 tok/s 72.7 tok/s 10.8 tok/s burst 58.6 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 ⓘ

$0.3779

$0.3051

$0.1194

$0.3037

$0.3049

缓存读取占比 ⓘ 20% 71% 81% 71% 71%

揭晓

通道

服务商 WorldBase.ai ↗

通道M-

官方基准

服务商 luminai ↗

通道M-Kiro/Antigravity

服务商 FastCode ↗

通道cc

服务商 ClaudeCN ↗

通道O-Max

返回列表