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

2026/06/25 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

FastCode

ClaudeAPI

O-Bdrk

ClaudeCN

O-Max

LingxiCode

M-

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 v2.1.160 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 04:57:10 ~ 04:57:50 (35.7s) 04:57:08 ~ 04:57:33 (22.0s) 04:57:06 ~ 04:57:11 (4.7s) 04:57:12 ~ 04:57:17 (3.8s) 04:57:14 ~ 04:57:24 (5.0s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 34,140 ms 19,697 ms 3,032 ms 2,212 ms 3,081 ms

首文本(SSE delta) 34,141 ms 19,698 ms 3,034 ms 2,213 ms 3,083 ms

总耗时 34,173 ms 20,579 ms 3,385 ms 2,710 ms 3,572 ms

输入 10 10 10 10 10

输出 54 35 40 58 38

缓存创建

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,123

5m: 0 / 1h: 33,123

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 0 0 0

输出速度 burst 39.7 tok/s 114.0 tok/s 116.7 tok/s 77.7 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 05:00:56 ~ 05:01:06 (7.6s) 05:01:01 ~ 05:01:11 (7.0s) 04:59:12 ~ 04:59:22 (6.3s) 04:59:47 ~ 04:59:57 (5.2s) 05:01:06 ~ 05:01:16 (6.2s)

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。自报身份，输出严格遵守以下 3 行格式（不允许 markdown、代码块、额外文本）：
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) 2,948 ms 4,541 ms 2,944 ms 2,672 ms 1,906 ms

首文本(SSE delta) 2,950 ms 4,542 ms 2,945 ms 2,673 ms 1,907 ms

总耗时 6,402 ms 5,420 ms 5,051 ms 4,134 ms 4,999 ms

输入 10 3 10 10 10

输出 179 26 229 181 212

缓存创建

192

5m: 0 / 1h: 192

1,678

5m: 0 / 1h: 1,678

234

5m: 0 / 1h: 234

252

5m: 0 / 1h: 252

232

5m: 0 / 1h: 232

缓存读取 33,122 31,487 33,123 33,122 33,122

输出速度 51.9 tok/s - 108.7 tok/s 123.9 tok/s 68.6 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-haiku-4-5-20251001

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

步骤 3: cutoff text

时间 05:02:34 ~ 05:02:44 (5.5s) 05:03:13 ~ 05:03:23 (5.4s) 05:03:15 ~ 05:03:25 (5.0s) 05:01:28 ~ 05:01:33 (4.0s) 05:02:18 ~ 05:02:23 (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) 3,356 ms 3,500 ms 3,013 ms 2,670 ms 2,247 ms

首文本(SSE delta) 3,357 ms 3,501 ms 3,014 ms 2,671 ms 2,250 ms

总耗时 4,287 ms 4,141 ms 3,559 ms 3,158 ms 3,460 ms

输入 10 3 10 10 10

输出 81 8 71 61 71

缓存创建

5m: 0 / 1h: 59

5m: 0 / 1h: 61

269

5m: 0 / 1h: 269

204

5m: 0 / 1h: 204

235

5m: 0 / 1h: 235

缓存读取 33,314 33,165 33,357 33,374 33,354

输出速度 87.1 tok/s - 130.3 tok/s 125.3 tok/s 58.7 tok/s

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 05:04:01 ~ 05:04:11 (5.8s) 05:06:08 ~ 05:06:18 (8.2s) 05:05:05 ~ 05:05:15 (7.1s) 05:04:51 ~ 05:05:01 (8.0s) 05:03:48 ~ 05:03:58 (9.1s)

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) 3,180 ms 6,313 ms 3,550 ms 2,750 ms 2,378 ms

首文本(SSE delta) 3,182 ms 6,314 ms 3,551 ms 2,750 ms 2,380 ms

总耗时 4,519 ms 6,896 ms 5,808 ms 6,967 ms 8,097 ms

输入 10 3 10 10 10

输出 104 56 344 775 397

缓存创建

5m: 0 / 1h: 35

33,261

5m: 0 / 1h: 33,261

113

5m: 0 / 1h: 113

5m: 0 / 1h: 95

109

5m: 0 / 1h: 109

缓存读取 33,373 0 33,626 33,578 33,589

输出速度 77.8 tok/s 96.2 tok/s 152.4 tok/s 183.8 tok/s 69.4 tok/s

响应

我是 Claude，由 Anthropic 公司开发的 AI 助手。我基于 Claude Haiku 4.5 模型，专注于帮助用户完成软件工程任务。

我是 Claude，由 Anthropic 公司开发的 AI 助手。我可以帮助用户完成软件工程任务、代码编写、问题解答等工作。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我可以帮助编程、研究、写作、分析等多种任务。

我是Claude，由Anthropic开发的AI助手。可以帮您进行编程、分析、问题解决等各种任务。

我是Claude，由Anthropic公司开发的AI助手。可以帮助编程、分析代码、回答问题和各类写作任务。

步骤 5: knowledge_recall text

时间 05:06:53 ~ 05:07:13 (19.3s) 05:08:37 ~ 05:08:47 (6.4s) 05:06:26 ~ 05:06:46 (18.3s) 05:08:51 ~ 05:09:11 (15.2s) 05:07:53 ~ 05:08:48 (51.0s)

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) 2,425 ms 3,588 ms 1,957 ms 2,110 ms 1,712 ms

首文本(SSE delta) 2,426 ms 3,590 ms 1,973 ms 2,111 ms 1,714 ms

总耗时 18,044 ms 4,984 ms 17,154 ms 14,172 ms 50,148 ms

输入 10 3 10 10 10

输出 1,342 191 2,552 1,998 4,183

缓存创建

8,078

5m: 0 / 1h: 8,078

8,076

5m: 0 / 1h: 8,076

8,701

5m: 0 / 1h: 8,701

9,066

5m: 0 / 1h: 9,066

8,713

5m: 0 / 1h: 8,713

缓存读取 0 0 0 0 0

输出速度 85.9 tok/s 137.0 tok/s 168.1 tok/s 165.7 tok/s 86.4 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

```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

```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

步骤 6: digit_count text

时间 05:10:23 ~ 05:10:43 (17.8s) 05:09:56 ~ 05:10:36 (37.8s) 05:09:48 ~ 05:10:23 (32.0s) 05:13:03 ~ 05:13:38 (31.4s) 05:09:54 ~ 05:10:14 (17.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=352638690664210687087249194122578412150662428027401432122459837562487862

HTTP 头(TTFB) 1,925 ms 18,536 ms 1,942 ms 2,298 ms 1,039 ms

首文本(SSE delta) 1,927 ms 18,537 ms 1,943 ms 2,299 ms 1,041 ms

总耗时 16,492 ms 36,308 ms 30,891 ms 30,650 ms 16,730 ms

输入 193 193 193 193 193

输出 1,205 2,745 6,240 5,266 3,171

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 82.7 tok/s 154.5 tok/s 215.6 tok/s 185.7 tok/s 202.1 tok/s

响应

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

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

```json
{"n": 73, "c": [6, 6, 4]}
```

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

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

总计

总输入(含缓存) 141,538 141,065 142,789 143,056 142,719

总输出 2,965 3,061 9,476 8,339 8,072

总缓存创建

41,486

5m: 0 / 1h: 41,486

76,198

5m: 0 / 1h: 76,198

42,440

5m: 0 / 1h: 42,440

42,739

5m: 0 / 1h: 42,739

42,411

5m: 0 / 1h: 42,411

总缓存读取 99,809 64,652 100,106 100,074 100,065

输出速度 81.1 tok/s 146.7 tok/s 191.9 tok/s 177.1 tok/s 108.2 tok/s

总执行时间 1m 31.6s 1m 27.0s 1m 13.4s 1m 7.6s 1m 33.6s

总墙钟时间

13m 33.1s

含步骤间等待

13m 28.0s

含步骤间等待

13m 17.4s

含步骤间等待

16m 26.1s

含步骤间等待

13m 0.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0769

Anthropic: Claude Haiku 4.5

$0.1172

Anthropic: Claude Haiku 4.5

$0.1107

Anthropic: Claude Haiku 4.5

$0.1054

Anthropic: Claude Haiku 4.5

$0.1036

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%) one-api (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 cloudflare -

ID 格式 msg_vrtx_01BvT47X3nXtnb6Gn5k5oU9z msg_01Dw6HEc3BwumgBfwoJSCUMX msg_01EmFaTe5oHNx16z7BtAoZia msg_016e7nehC96KjutswcddUDfL msg_01UkxGYZwaW5qfoqAArcQsMj

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
cc	server: nginx	id_format: msg_prefix extra_fields: ["stop_details"] has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	body:id:msg_prefixbody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
O-Bdrk	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606242109579787083208268d9d6pLnaMB9K	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-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—
O-Max	cf-ray: a10eb73bbf13df76-AMS server: cloudflare x-oneapi-request-id: 20260625050950603655108yYIDFVWg	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-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	a10eb73bbf13df76-AMS
M-	cf-ray: a10ebbf98f54e393-NRT server: cloudflare set-cookie: server_name_session=3de609b8cf2a7e28d141c3b6c8b41f57; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 202606242113048341718982THoHVOu	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-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	a10ebbf98f54e393-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_011CcNmeBiNZ5Y9mtose8SNP

协议指纹评分（机器自动）

总分 / 100 82 85 98 97 100

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 5.1 /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 7.7 /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 10.0 /10 ×8 10.0 /10 ×8 10.0 /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 0.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6

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

延迟基线 10.0 /10 ×5 7.5 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

Req-ID 透传 0.0 /10 ×4 0.0 /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 82 v3.24.1 85 v3.24.1 98 v3.24.1 97 v3.24.1 基准（参考）

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 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 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 13986 ms 13055 ms 10975 ms 10299 ms 14501 ms

输出速度 81.1 tok/s 146.7 tok/s 191.9 tok/s 177.1 tok/s 108.2 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 ⓘ

$0.0769

$0.1172

$0.1107

$0.1054

$0.1036

缓存读取占比 ⓘ 71% 46% 70% 70% 70%

揭晓

通道

服务商 FastCode ↗

通道cc

服务商 ClaudeAPI ↗

通道O-Bdrk

服务商 ClaudeCN ↗

通道O-Max

服务商 LingxiCode ↗

通道M-

官方基准

返回列表