模型对比 | RPDiag

测试说明点击展开 5 步 quick-probe-v1 详解

前 4 步同一会话（交互 TUI 多轮），第 5、6 步各自独立新会话，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 1-3 步的延迟在交互进程内发生，让缓存前缀在墙钟里老化）。第 3 步（cutoff）累计跨度 > 5 分钟保住 sliding cache 检测；总跨度 > 6 分钟。第 5 步（知识召回）与第 6 步（档位）都是独立新会话（第 6 步不复用上下文，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

2026/07/02 运行日期

claude-haiku-4-5-20251001 官方基准

4 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

ByteCatCode

O-Max

WorldBase.ai

M-

AnyAIGC

M-

FoxCode

R-super

官方基准

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 v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195

偏差标红基于官方基准

步骤 1: ping text

时间 22:56:06 ~ 23:06:07 (598.6s) 22:56:04 ~ 23:06:10 (604.1s) 22:56:04 ~ 23:05:05 (536.8s) 22:56:08 ~ 23:04:53 (521.4s) 22:56:10 ~ 23:03:41 (450.0s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 4,133 ms 3,544 ms 3,179 ms 2,148 ms 1,875 ms

首文本(SSE delta) - 3,560 ms 3,180 ms 2,149 ms 2,281 ms

总耗时 4,396 ms 4,140 ms 3,181 ms 3,141 ms 2,458 ms

输入 362 3,497 9,632 22,833 10

输出 11 57 2 102 47

缓存创建

25,079

5m: 25,079 / 1h: 0

19,815

5m: 19,815 / 1h: 0

5m: 0 / 1h: 0

6,392

5m: 0 / 1h: 6,392

缓存读取 0 0 0 0 16,545

输出速度 - 98.3 tok/s - 102.8 tok/s -

响应

{"title": "pong"}

pong

pong

pong

pong

步骤 2: identity text

时间 22:56:06 ~ 23:06:07 (598.6s) 22:56:04 ~ 23:06:10 (604.1s) 22:56:04 ~ 23:05:05 (536.8s) 22:56:08 ~ 23:04:53 (521.4s) 22:56:10 ~ 23:03:41 (450.0s)

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) 5,852 ms 7,158 ms 4,683 ms 1,984 ms 1,955 ms

首文本(SSE delta) - 7,160 ms 4,684 ms 1,985 ms 2,756 ms

总耗时 8,137 ms 9,905 ms 5,352 ms 3,667 ms 5,878 ms

输入 10 3,528 9,822 23,447 10

输出 204 208 25 211 223

缓存创建

5m: 0 / 1h: 0

917

5m: 917 / 1h: 0

5m: 0 / 1h: 0

231

5m: 0 / 1h: 231

缓存读取 17,922 19,070 0 0 22,937

输出速度 89.3 tok/s 75.8 tok/s - 125.4 tok/s 71.4 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-haiku-4-5-20251001

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

步骤 3: cutoff text

时间 22:56:06 ~ 23:06:07 (598.6s) 22:56:04 ~ 23:06:10 (604.1s) 22:56:04 ~ 23:05:05 (536.8s) 22:56:08 ~ 23:04:53 (521.4s) 22:56:10 ~ 23:03:41 (450.0s)

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) 3,158 ms 3,108 ms 9,193 ms 2,457 ms 2,276 ms

首文本(SSE delta) - 3,109 ms 9,194 ms 2,458 ms 2,912 ms

总耗时 4,363 ms 4,013 ms 9,885 ms 3,331 ms 2,976 ms

输入 10 3,534 9,879 24,010 10

输出 104 65 5 106 72

缓存创建

243

5m: 0 / 1h: 243

983

5m: 983 / 1h: 0

5m: 0 / 1h: 0

263

5m: 0 / 1h: 263

缓存读取 17,743 19,039 0 0 23,168

输出速度 86.3 tok/s 71.9 tok/s - 121.4 tok/s -

响应

2025-02

2025-02

2025-05

2025-02

2025-02

步骤 4: identity_free text

时间 22:56:06 ~ 23:06:07 (598.6s) 22:56:04 ~ 23:06:10 (604.1s) 22:56:04 ~ 23:05:05 (536.8s) 22:56:08 ~ 23:04:53 (521.4s) 22:56:10 ~ 23:03:41 (450.0s)

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) 2,731 ms 4,656 ms 5,521 ms 1,889 ms 5,819 ms

首文本(SSE delta) - 4,676 ms 5,522 ms 1,890 ms 6,726 ms

总耗时 10,677 ms 7,851 ms 5,531 ms 8,708 ms 11,186 ms

输入 10 3,574 9,916 24,475 10

输出 752 344 27 1,837 445

缓存创建

612

5m: 0 / 1h: 612

20,248

5m: 20,248 / 1h: 0

5m: 0 / 1h: 0

114

5m: 0 / 1h: 114

缓存读取 17,743 0 0 0 23,431

输出速度 94.6 tok/s 108.3 tok/s - burst 99.8 tok/s

响应

ææ¯Claudeï¼ç±Anthropicå¼åçAIå©æï¼å¯å¸®å©ç¨æ·å¤çææ¬ãä»£ç ååæçä»»å¡ã

ä»£ç ç¼åãé®é¢åæçä»»å¡åå©ã

ææ¯ç±anthropicæé çAIå©æï¼åä¸ºClaudeï¼ä¸»è¦å¸®å©ç¨æ·å®æä¿¡æ¯æ¥è¯¢ãä»£ç ç¼åçä»»å¡ã

ææ¯Claude Haiku 4.5ï¼ç±Anthropicæé ãå¸®å©ä»£ç ç¼åãè°è¯ãåä½åç ç©¶çå·¥ä½ã

å®æ¹AIç¼ç¨å©æãåºäºClaude Haiku 4.5æ¨¡åï¼å¯ä½ä¸ºCLIãæ¡é¢åºç¨ãç½é¡µåºç¨åIDEæ©å±ä½¿ç¨ï¼å¸®å©ç¨æ·è¿è¡è½¯ä»¶å·¥ç¨ä»»å¡ã

步骤 5: knowledge_recall text

时间 23:14:46 ~ 23:15:36 (49.7s) 23:14:58 ~ 23:15:18 (19.6s) 23:14:37 ~ 23:14:57 (17.7s) 23:14:16 ~ 23:14:37 (17.9s) 23:13:05 ~ 23:17:20 (34.2s)

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) 2,404 ms 2,006 ms 3,630 ms 1,367 ms 1,340 ms

首文本(SSE delta) - 2,007 ms 3,632 ms 1,370 ms 2,165 ms

总耗时 38,843 ms 7,931 ms 4,333 ms 7,219 ms 24,195 ms

输入 10 1,146 8,157 7,095 10

输出 4,060 880 154 993 2,509

缓存创建

28,951

5m: 0 / 1h: 28,951

6,491

5m: 6,491 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 0 0 6,815

输出速度 111.4 tok/s 148.5 tok/s 219.7 tok/s 169.8 tok/s 113.9 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":"UNKNOWN","oscars2024_best_picture":"UNKNOWN","ucl_2024":"UNKNOWN","nba_2024":"UNKNOWN","copa_america_2024":"UNKNOWN","euro2024":"UNKNOWN","nobel_peace_2024":"UNKNOWN","wo

```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

步骤 6: digit_count text

时间 23:18:52 ~ 23:19:27 (31.1s) 23:18:09 ~ 23:18:24 (11.4s) 23:17:24 ~ 23:17:34 (5.9s) 23:16:44 ~ 23:16:59 (13.7s) 23:19:15 ~ 23:19:40 (22.4s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=338698803809761841193158771990393188734148909343726557788051010089343286

HTTP 头(TTFB) 2,044 ms 2,023 ms 2,096 ms 1,405 ms 1,745 ms

首文本(SSE delta) - 2,039 ms 2,097 ms 1,406 ms 1,766 ms

总耗时 28,881 ms 10,336 ms 2,098 ms 12,770 ms 21,414 ms

输入 193 66 165 210 193

输出 3,265 2,297 14 2,699 2,510

缓存创建

5m: 0 / 1h: 0

146

5m: 146 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 582 0 0 0

输出速度 121.7 tok/s burst - 237.5 tok/s 127.7 tok/s

响应

```json
{"n":72,"c":[8,7,9]}
```

```json
{"n":72,"c":[8,7,9]}
```

{"n":66,"c":[5,4,4]}

```json
{"n":72,"c":[8,7,9]}
```

```json
{"n":72,"c":[8,7,9]}
```

总计

总输入(含缓存) 108,888 102,636 47,571 102,070 100,139

总输出 8,396 3,851 227 5,948 5,806

总缓存创建

54,885

5m: 25,079 / 1h: 29,806

48,600

5m: 48,600 / 1h: 0

5m: 0 / 1h: 0

7,000

5m: 0 / 1h: 7,000

总缓存读取 53,408 38,691 0 0 92,896

输出速度 112.2 tok/s 116.6 tok/s 219.7 tok/s 198.0 tok/s 115.4 tok/s

总执行时间 41m 15.2s 40m 47.4s 36m 10.8s 35m 17.3s 30m 56.5s

总墙钟时间

23m 21.5s

含步骤间等待

22m 20.5s

含步骤间等待

21m 30.6s

含步骤间等待

20m 51.3s

含步骤间等待

23m 29.9s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1048

Anthropic: Claude Haiku 4.5

$0.0764

Anthropic: Claude Haiku 4.5

$0.0487

Anthropic: Claude Haiku 4.5

$0.1318

Anthropic: Claude Haiku 4.5

$0.0473

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 one-api (100%) direct (100%) direct (50%) one-api (100%) direct (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_01CBt7JidijSoZEte1HXbLGW msg_01SUmCfeLiBsorEiHB2gPVHT 自定义 msg_01rKsTt2tW0jQO90TETcAr9P msg_01VEWwM69ZZjJMvHS21zsL7i

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/1.29.8 x-oneapi-request-id: 202607021518546788268228268d9d6dEfbrFi8	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-	cf-ray: a14e9f1c9e636053-AMS server: cloudflare	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: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	1addea6f-dcb6-4937-96b3-d2bcf8a32abd a14e9f1c9e636053-AMS
M-	server: nginx	id_format: non_standard has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	body:id:non_standardbody:model:date_suffixbody:usage.cache_fields	202607021517289300722778268d9d6LQQ25LZf
R-super	cf-ray: a14e9d063853e393-NRT server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202607021516453207483648268d9d6spKan0Oj	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	a14e9d063853e393-NRT
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_011CcdTNc5bP8gt8J9zvpEqj

协议指纹评分（机器自动）

总分 / 100 82 68 36 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已对照基线

缓存命中比 7.5 /10 ×20 0.0 /10 ×20 0.0 /10 ×20 0.0 /10 ×20 10.0 /10 ×20

缓存 TTL 一致 7.0 /10 ×15 0.0 /10 ×15 0.0 /10 ×15 0.0 /10 ×15 10.0 /10 ×15

延迟基线 8.8 /10 ×15 9.0 /10 ×15 8.2 /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 5.0 /10 ×14 5.0 /10 ×14 10.0 /10 ×14

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

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

系统提示纯净 0.0 /10 ×8 5.0 /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

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 0.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

知识截止 10.0 /10 ×5 10.0 /10 ×5 5.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

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

stop_reason 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3

Req-ID 透传 0.0 /10 ×2 0.0 /10 ×2 0.0 /10 ×2 0.0 /10 ×2 10.0 /10 ×2

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 82 v3.27.0 68 v3.27.0 36 v3.27.0 69 v3.27.0 基准（参考）

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 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 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 15883 ms 7363 ms 5063 ms 6473 ms 11351 ms

输出速度 112.2 tok/s 116.6 tok/s 219.7 tok/s 198.0 tok/s 115.4 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 ⓘ

$0.1048

$0.0764

$0.0487

$0.1318

$0.0473

缓存读取占比 ⓘ 49% 38% 0% 0% 93%

揭晓

通道

服务商 ByteCatCode ↗

通道O-Max

服务商 WorldBase.ai ↗

通道M-

服务商 AnyAIGC ↗

通道M-

服务商 FoxCode ↗

通道R-super

官方基准

返回列表