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

2026/06/21 运行日期

claude-haiku-4-5-20251001 官方基准

1 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

TopRouterCN

O-Max

官方基准

Novaris

O-Max

AIMZ

M-Max

SAIAi

O-Max

基本信息

服务名 Claude Code Claude Code Claude Code Claude Code Claude Code

模型 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001 claude-haiku-4-5-20251001

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 16:16:30 ~ 16:16:40 (5.4s) 16:16:34 ~ 16:16:39 (3.1s) 16:16:28 ~ 16:16:33 (4.0s) 16:16:26 ~ 16:16:31 (4.2s) 16:16:32 ~ 16:16:37 (4.4s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 4,359 ms 1,480 ms 2,523 ms 2,452 ms 2,703 ms

首文本(SSE delta) 4,360 ms 1,482 ms 2,524 ms 2,657 ms 2,705 ms

总耗时 4,451 ms 1,866 ms 3,001 ms 2,736 ms 3,267 ms

输入 1,645 10 10 10 10

输出 53 49 63 41 45

缓存创建

31,487

5m: 0 / 1h: 31,487

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

32,785

5m: 0 / 1h: 32,785

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 0 0 0

输出速度 - 127.6 tok/s 132.1 tok/s - 80.1 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 16:19:03 ~ 16:19:13 (6.7s) 16:18:47 ~ 16:18:52 (4.6s) 16:19:49 ~ 16:19:59 (5.9s) 16:20:23 ~ 16:20:33 (9.8s) 16:19:57 ~ 16:20:07 (5.9s)

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) 5,885 ms 1,646 ms 1,888 ms 2,451 ms 2,403 ms

首文本(SSE delta) 5,886 ms 1,648 ms 1,889 ms 8,290 ms 2,404 ms

总耗时 5,953 ms 3,500 ms 5,059 ms 8,554 ms 4,915 ms

输入 10 10 10 10 10

输出 247 197 387 549 250

缓存创建

1,879

5m: 0 / 1h: 1,879

240

5m: 0 / 1h: 240

254

5m: 0 / 1h: 254

232

5m: 0 / 1h: 232

236

5m: 0 / 1h: 236

缓存读取 31,487 33,122 33,122 32,785 33,122

输出速度 burst 106.4 tok/s 122.1 tok/s burst 99.6 tok/s

响应

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

vendor: anthropic
brand: claude
model: claude-3-5-sonnet

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

步骤 3: cutoff text

时间 16:22:43 ~ 16:22:53 (5.7s) 16:21:44 ~ 16:21:49 (4.2s) 16:21:54 ~ 16:22:04 (5.7s) 16:22:25 ~ 16:22:30 (3.8s) 16:24:06 ~ 16:24:16 (5.6s)

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) 4,801 ms 2,637 ms 3,889 ms 2,130 ms 3,742 ms

首文本(SSE delta) 4,802 ms 2,638 ms 3,890 ms 2,482 ms 3,745 ms

总耗时 4,883 ms 3,157 ms 4,828 ms 2,757 ms 4,632 ms

输入 10 10 10 10 10

输出 65 64 140 71 66

缓存创建

287

5m: 0 / 1h: 287

237

5m: 0 / 1h: 237

429

5m: 0 / 1h: 429

591

5m: 0 / 1h: 591

273

5m: 0 / 1h: 273

缓存读取 33,366 33,362 33,376 33,017 33,358

输出速度 - 123.3 tok/s 149.3 tok/s - 74.4 tok/s

响应

2025-02

2025-02

2025-02

2024-04

2025-02

步骤 4: identity_free text

时间 16:24:56 ~ 16:25:06 (7.2s) 16:22:52 ~ 16:23:02 (8.5s) 16:24:53 ~ 16:25:04 (7.4s) 16:26:30 ~ 16:26:41 (6.6s) 16:25:53 ~ 16:26:04 (6.5s)

提示词

查看本步 Prompt

请用中文介绍你自己是什么产品或助手，由谁开发，50 字以内。

HTTP 头(TTFB) 6,145 ms 5,562 ms 2,013 ms 2,290 ms 1,882 ms

首文本(SSE delta) 6,146 ms 5,576 ms 2,014 ms 4,016 ms 1,884 ms

总耗时 6,226 ms 7,518 ms 6,508 ms 4,265 ms 5,559 ms

输入 10 10 10 10 10

输出 576 342 672 328 340

缓存创建

107

5m: 0 / 1h: 107

106

5m: 0 / 1h: 106

182

5m: 0 / 1h: 182

113

5m: 0 / 1h: 113

100

5m: 0 / 1h: 100

缓存读取 33,653 33,599 33,805 33,608 33,631

输出速度 burst 176.1 tok/s 149.5 tok/s burst 92.5 tok/s

响应

我是Claude，由Anthropic开发的AI助手。专长于软件工程任务，包括代码编写、调试、审查和重构等工作。

我是Claude，Anthropic公司开发的AI助手。我可以帮助用户进行编程、分析、创意写作等多项任务。

我是Claude，由Anthropic公司开发的AI助手。能帮助编程、分析、写作和解答问题。我追求安全、诚实和有益的服务。

我是Claude，由Anthropic公司开发的AI助手。我能帮助用户进行代码编辑、分析、问题解决等各种软件工程任务。

我是Claude，由Anthropic公司开发的AI助手。我具有多模态能力，可以帮助编程、写作、分析等各种任务。

步骤 5: knowledge_recall text

时间 16:26:25 ~ 16:26:40 (12.1s) 16:26:46 ~ 16:27:01 (13.4s) 16:28:28 ~ 16:28:43 (14.3s) 16:29:45 ~ 16:30:00 (10.8s) 16:29:47 ~ 16:30:42 (50.8s)

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) 7,110 ms 1,603 ms 1,639 ms 1,651 ms 1,571 ms

首文本(SSE delta) 7,111 ms 1,606 ms 1,640 ms 8,270 ms 1,573 ms

总耗时 11,396 ms 12,363 ms 13,427 ms 9,702 ms 49,607 ms

输入 10 10 10 10 10

输出 1,281 1,787 2,018 1,058 4,560

缓存创建

8,954

5m: 0 / 1h: 8,954

8,665

5m: 0 / 1h: 8,665

9,277

5m: 0 / 1h: 9,277

8,667

5m: 0 / 1h: 8,667

8,689

5m: 0 / 1h: 8,689

缓存读取 0 0 0 0 0

输出速度 burst 166.1 tok/s 171.2 tok/s burst 94.9 tok/s

响应

```json
{"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

```json
{"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

{"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

```json
{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"UNKNOWN","nba_2024":"UNKNOWN","copa_america_2024":"UNKNOWN","euro2024":"UNKNOWN","nobel_pe

{"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 16:28:21 ~ 16:28:46 (23.3s) 16:29:56 ~ 16:30:16 (16.4s) 16:31:04 ~ 16:31:19 (13.2s) 16:33:08 ~ 16:33:38 (27.8s) 16:34:14 ~ 16:34:39 (24.0s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=159320992976270746600134974472485352610435909278612759390615140526765854

HTTP 头(TTFB) 8,499 ms 1,235 ms 1,098 ms 1,700 ms 2,054 ms

首文本(SSE delta) 8,500 ms 1,261 ms 1,099 ms 1,702 ms 2,079 ms

总耗时 22,530 ms 15,495 ms 12,386 ms 25,615 ms 22,917 ms

输入 193 193 193 139 193

输出 3,495 2,919 2,029 4,774 2,721

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 249.1 tok/s 205.1 tok/s 179.8 tok/s 199.6 tok/s 130.6 tok/s

响应

```json
{"n":72,"c":[8,8,9]}
```

```json
{"n": 72, "c": [8, 8, 9]}
```

```json
{"n":72,"c":[8,8,9]}
```

```json
{"n":72,"c":[8,8,9]}
```

```json
{"n":72,"c":[8,8,9]}
```

总计

总输入(含缓存) 143,098 142,696 143,810 141,987 142,774

总输出 5,717 5,358 5,309 6,821 7,982

总缓存创建

42,714

5m: 0 / 1h: 42,714

42,370

5m: 0 / 1h: 42,370

43,264

5m: 0 / 1h: 43,264

42,388

5m: 0 / 1h: 42,388

42,420

5m: 0 / 1h: 42,420

总缓存读取 98,506 100,083 100,303 99,410 100,111

输出速度 249.1 tok/s 180.5 tok/s 165.1 tok/s 199.6 tok/s 104.3 tok/s

总执行时间 1m 0.5s 50.1s 50.4s 1m 3.0s 1m 37.2s

总墙钟时间

12m 16.2s

含步骤间等待

13m 42.0s

含步骤间等待

14m 51.0s

含步骤间等待

17m 12.2s

含步骤间等待

18m 7.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0937

Anthropic: Claude Haiku 4.5

$0.0900

Anthropic: Claude Haiku 4.5

$0.0909

Anthropic: Claude Haiku 4.5

$0.0972

Anthropic: Claude Haiku 4.5

$0.1032

Anthropic: Claude Haiku 4.5

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%) one-api (100%) direct (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_013q5JzqACUz6AxfN58n9Ci8 msg_016f8eYWMUBTAoRQGabGpLPF msg_01DvZDLVS4LUND2FReHoX7DX msg_01Sg52ufFC1KSGJ75ZVrfKRS msg_01TPpbDWVfMHRehipXkYnLij

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/1.27.5	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	tr-ew7y4grd-bnfxa1qgw8-127228
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: true usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcG6GCLttr38eDM8AeHmt
O-Max	server: nginx x-oneapi-request-id: 202606210831051369372138268d9d6eBNcF8pj	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-oneapi-request-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
M-Max	cf-ray: a0f1aab9cdc933b1-AMS server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606210833108892952718268d9d6ZHVtL3cD	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: true has_inference_geo: true model_has_date_suffix: true 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:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a0f1aab9cdc933b1-AMS
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: true usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcG6bG5f11NBqfWuur3BM

协议指纹评分（机器自动）

总分 / 100 98 100 98 81 100

基线状态已对照基线已对照基线已对照基线已对照基线已对照基线

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

缓存 TTL 一致 10.0 /10 ×15 10.0 /10 ×15 10.0 /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 10.0 /10 ×14 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 10.0 /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 10.0 /10 ×8 10.0 /10 ×8

系统提示纯净 10.0 /10 ×8 10.0 /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 6.7 /10 ×7 10.0 /10 ×7

知识截止 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 0.0 /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 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

延迟基线 9.3 /10 ×5 10.0 /10 ×5 10.0 /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 10.0 /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

流式投递未计入未计入未计入未计入未计入

综合结论

与基线相似度 98 v3.24.0 基准（参考） 98 v3.24.0 81 v3.24.0 100 v3.24.0

雷达图例

共 23 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 自由身份 identity_free_clean
6 停止原因 stop_reason_present
7 系统提示纯净 system_prompt_clean
8 Service Tier service_tier_present
9 缓存命中比 cache_hit_ratio_match
10 缓存 TTL 一致 cache_ttl_consistency
11 推理地区 inference_geo_present
12 Thinking 体量 thinking_volume_match
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 自报身份一致 self_identity_consistency
20 信封自报一致 envelope_self_report_match
21 中文遵循 instruction_following_lang
22 档位思考量 tier_thinking_volume_match
23 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 9240 ms 7317 ms 7535 ms 8938 ms 15150 ms

输出速度 249.1 tok/s 180.5 tok/s 165.1 tok/s 199.6 tok/s 104.3 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 ⓘ

$0.0937

$0.0900

$0.0909

$0.0972

$0.1032

缓存读取占比 ⓘ 69% 70% 70% 70% 70%

揭晓

通道

服务商 TopRouterCN ↗

通道O-Max

官方基准

服务商 Novaris ↗

通道O-Max

服务商 AIMZ ↗

通道M-Max

服务商 SAIAi ↗

通道O-Max

返回列表