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-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

SSSAiCode

R-Kiro

官方基准

ClaudeAPI

O-Bdrk

SAIAi

O-Max

PackyCode

cc-Zeus

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 16:22:48 ~ 16:22:53 (4.1s) 16:22:52 ~ 16:22:58 (4.6s) 16:22:46 ~ 16:22:52 (4.7s) 16:22:51 ~ 16:22:56 (4.6s) 16:22:44 ~ 16:22:50 (3.4s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,662 ms 3,091 ms 2,333 ms 2,698 ms 1,753 ms

首文本(SSE delta) 2,686 ms 3,093 ms 2,762 ms 2,701 ms 1,754 ms

总耗时 3,115 ms 3,432 ms 2,890 ms 3,465 ms 2,280 ms

输入 0 10 10 10 10

输出 51 47 44 53 44

缓存创建

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

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 118.9 tok/s 138.6 tok/s - 69.4 tok/s 83.7 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 16:24:00 ~ 16:24:05 (4.9s) 16:25:19 ~ 16:25:24 (4.3s) 16:26:19 ~ 16:26:29 (5.6s) 16:24:44 ~ 16:24:54 (5.4s) 16:24:23 ~ 16:24:33 (5.4s)

提示词

查看本步 Prompt

以下是对你自身的 metadata 探针。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。按以下 3 行回答自己的身份，每行一个 label，不要寒暄不要解释：
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) 2,524 ms 2,039 ms 1,913 ms 1,678 ms 1,616 ms

首文本(SSE delta) 2,526 ms 2,041 ms 2,258 ms 1,680 ms 1,617 ms

总耗时 4,073 ms 3,364 ms 3,963 ms 4,439 ms 3,716 ms

输入 0 10 10 10 10

输出 210 177 235 182 248

缓存创建

251

5m: 0 / 1h: 251

248

5m: 0 / 1h: 248

257

5m: 0 / 1h: 257

248

5m: 0 / 1h: 248

缓存读取 0 33,122 33,122 33,122 33,122

输出速度 135.7 tok/s 133.8 tok/s 137.8 tok/s 66.0 tok/s 118.2 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 16:27:02 ~ 16:27:07 (3.5s) 16:28:06 ~ 16:28:11 (4.2s) 16:28:18 ~ 16:28:23 (4.2s) 16:26:00 ~ 16:26:05 (3.7s) 16:26:44 ~ 16:26:49 (3.7s)

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) 2,211 ms 2,727 ms 2,111 ms 1,842 ms 2,320 ms

首文本(SSE delta) 2,212 ms 2,728 ms 2,702 ms 1,844 ms 2,321 ms

总耗时 2,740 ms 3,178 ms 2,820 ms 2,726 ms 2,489 ms

输入 0 10 10 10 10

输出 75 70 73 61 83

缓存创建

200

5m: 0 / 1h: 200

277

5m: 0 / 1h: 277

205

5m: 0 / 1h: 205

290

5m: 0 / 1h: 290

缓存读取 0 33,373 33,370 33,379 33,370

输出速度 142.0 tok/s 155.6 tok/s - 69.2 tok/s -

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 16:30:29 ~ 16:30:39 (4.9s) 16:30:41 ~ 16:30:51 (5.3s) 16:30:26 ~ 16:30:36 (5.3s) 16:28:30 ~ 16:28:40 (6.1s) 16:30:11 ~ 16:30:16 (4.4s)

提示词

查看本步 Prompt

中文做个自我介绍：你是什么产品或助手？谁开发的？50 字内。

HTTP 头(TTFB) 2,417 ms 2,163 ms 1,929 ms 2,402 ms 1,994 ms

首文本(SSE delta) 2,452 ms 2,168 ms 1,930 ms 2,403 ms 1,995 ms

总耗时 4,035 ms 4,318 ms 4,050 ms 5,111 ms 3,345 ms

输入 0 10 10 10 10

输出 175 372 340 202 211

缓存创建

112

5m: 0 / 1h: 112

107

5m: 0 / 1h: 107

5m: 0 / 1h: 95

125

5m: 0 / 1h: 125

缓存读取 0 33,573 33,647 33,584 33,660

输出速度 110.5 tok/s 173.0 tok/s 160.4 tok/s 74.6 tok/s 156.3 tok/s

响应

我是 Claude 代理，由 Anthropic 开发。在 Claude Code 环境中作为交互式助手，我帮助用户完成软件工程任务，如编写代码、调试问题、代码审查等。

我是Claude Haiku，由Anthropic开发的AI助手。我可以帮助编程、分析、写作、回答问题等任务，具有良好的推理能力和代码理解能力。

我是Claude，一个由Anthropic公司开发的AI助手。我可以帮助编程、分析、写作等多种任务。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我能帮助用户完成软件工程任务、代码编写、问题解答等多种工作。

我是Claude，由Anthropic公司开发的AI助手。我可以帮助编写和调试代码、分析问题、进行研究和学习等多种任务。

步骤 5: knowledge_recall text

时间 16:32:16 ~ 16:32:31 (13.4s) 16:33:33 ~ 16:33:53 (15.2s) 16:34:15 ~ 16:34:35 (15.5s) 16:30:49 ~ 16:32:04 (70.0s) 16:31:53 ~ 16:32:13 (19.9s)

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) 1,787 ms 2,670 ms 1,576 ms 1,404 ms 1,504 ms

首文本(SSE delta) 1,811 ms 2,672 ms 1,577 ms 1,405 ms 1,505 ms

总耗时 11,367 ms 14,159 ms 13,910 ms 69,050 ms 18,596 ms

输入 0 10 10 10 10

输出 2,086 1,794 1,918 5,618 2,711

缓存创建

8,675

5m: 0 / 1h: 8,675

8,712

5m: 0 / 1h: 8,712

8,499

5m: 0 / 1h: 8,499

8,614

5m: 0 / 1h: 8,614

缓存读取 0 0 0 0 0

输出速度 218.3 tok/s 156.2 tok/s 155.5 tok/s 83.1 tok/s 158.6 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

```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_20

```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 16:35:45 ~ 16:36:00 (13.2s) 16:36:37 ~ 16:37:12 (33.3s) 16:36:31 ~ 16:36:56 (23.9s) 16:33:50 ~ 16:34:20 (27.2s) 16:34:47 ~ 16:35:22 (30.4s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=915802807542041455471886413019793191950944849161448708340686292125710327

HTTP 头(TTFB) 1,339 ms 1,107 ms 819 ms 1,547 ms 1,031 ms

首文本(SSE delta) 1,352 ms 1,109 ms 820 ms 1,549 ms 1,032 ms

总耗时 12,341 ms 32,235 ms 22,205 ms 26,238 ms 28,972 ms

输入 0 193 193 193 193

输出 1,891 2,880 4,181 4,949 5,841

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 172.1 tok/s 92.5 tok/s 195.5 tok/s 200.5 tok/s 209.1 tok/s

响应

```json
{"n":72,"c":[8,6,8]}
```

```json
{"n":72,"c":[8,6,8]}
```

```json
{"n":72,"c":[8,6,8]}
```

```json
{"n":72,"c":[8,6,8]}
```

```json
{"n":72,"c":[8,6,8]}
```

总计

总输入(含缓存) 0 142,671 142,848 142,506 142,794

总输出 4,488 5,340 6,791 11,065 9,138

总缓存创建

42,360

5m: 0 / 1h: 42,360

42,466

5m: 0 / 1h: 42,466

42,178

5m: 0 / 1h: 42,178

42,399

5m: 0 / 1h: 42,399

总缓存读取 0 100,068 100,139 100,085 100,152

输出速度 182.2 tok/s 113.9 tok/s 177.8 tok/s 111.3 tok/s 184.8 tok/s

总执行时间 43.9s 1m 6.9s 59.2s 1m 57.1s 1m 7.1s

总墙钟时间

13m 12.0s

含步骤间等待

14m 20.0s

含步骤间等待

14m 10.0s

含步骤间等待

11m 29.3s

含步骤间等待

12m 38.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

>=$0.0224

Anthropic: Claude Haiku 4.5 (成本不完整)

$0.0899

Anthropic: Claude Haiku 4.5

$0.0973

Anthropic: Claude Haiku 4.5

$0.1183

Anthropic: Claude Haiku 4.5

$0.1089

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 (75%) direct (100%) one-api (100%) direct (100%) one-api (100%)

上游 Anthropic 直连 (75%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%)

CDN - - - - cloudflare

ID 格式 msg_4b7c347113b948c786e3e1894f6ecaf5 msg_01PubvnCLn6jBZjQze5dTtJF msg_01PQXjTeNSa4raGfVRNgaqbS msg_01R4CrPxBBJTpEXCoMQQXZnZ msg_012rt62tnghgHVPSJV542hX7

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
R-Kiro	server: Photon-Edge x-amzn-requestid: e9d90640-4eda-4c20-8b33-dad69dd50707	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.cache_fieldsupstream:x-amzn-requestidupstream:anthropic-direct-from-body	1781858147017-xivwu8rhg478p e9d90640-4eda-4c20-8b33-dad69dd50707
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_011CcCK9ypfGbEZB2pS7eBS4
O-Bdrk	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606190836336405530398268d9d6bUnHU4xS	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	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_011CcCJwc86xa8Exaao5jDya
cc-Zeus	via: 1.1 Caddy cf-ray: a0e132627fd25730-LAX server: cloudflare x-oneapi-request-id: 20260619163449375413721XXNIy7Sf	id_format: msg_prefix via_chain: ["1.1 Caddy"] 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-rayhdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a0e132627fd25730-LAX

协议指纹评分（机器自动）

总分 / 100 56 100 98 100 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 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 5.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

sliding 5m 5.0 /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

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 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 10.0 /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

系统提示纯净未计入 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8

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 56 v3.24.0 基准（参考） 98 v3.24.0 100 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 自由身份 identity_free_clean
6 停止原因 stop_reason_present
7 Service Tier service_tier_present
8 缓存命中比 cache_hit_ratio_match
9 缓存 TTL 一致 cache_ttl_consistency
10 推理地区 inference_geo_present
11 Thinking 体量 thinking_volume_match
12 缓存连续 cache_continuity_intra
13 知识召回 knowledge_recall_match
14 延迟基线 latency_baseline_match
15 原生 msg-ID anthropic_msg_id_format
16 Sliding 缓存 cache_sliding_correctness
17 结构化身份 identity_structured_match
18 自报身份一致 self_identity_consistency
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 6279 ms 10114 ms 8306 ms 18505 ms 9900 ms

输出速度 182.2 tok/s 113.9 tok/s 177.8 tok/s 111.3 tok/s 184.8 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 ⓘ

>=$0.0224 (成本不完整)

$0.0899

$0.0973

$0.1183

$0.1089

缓存读取占比 ⓘ - 70% 70% 70% 70%

揭晓

通道

服务商 SSSAiCode ↗

通道R-Kiro

官方基准

服务商 ClaudeAPI ↗

通道O-Bdrk

服务商 SAIAi ↗

通道O-Max

服务商 PackyCode ↗

通道cc-Zeus

返回列表