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

2026/06/23 运行日期

claude-haiku-4-5-20251001 官方基准

1 通道分数 < 70 另含 1 任务失败

已揭晓揭晓进度

步骤 / 指标

duckcoding

cc专用

ClaudeAPI

O-Bdrk

官方基准

ByteCatCode

O-Max

FoxCode

R-ultra

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 00:43:14 ~ 00:43:19 (4.4s) 00:43:12 ~ 00:43:17 (4.8s) 00:43:16 ~ 00:43:21 (4.1s) 00:43:10 ~ 00:43:20 (5.7s) 00:43:08 ~ 00:43:13 (4.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,418 ms 2,892 ms 2,435 ms 3,282 ms 2,236 ms

首文本(SSE delta) 2,420 ms 2,898 ms 2,437 ms - 2,238 ms

总耗时 3,438 ms 3,414 ms 2,877 ms 3,453 ms 3,016 ms

输入 10 10 10 10 33,019

输出 61 51 56 53 67

缓存创建

33,122

5m: 0 / 1h: 33,122

32,406

5m: 32,406 / 1h: 0

33,122

5m: 0 / 1h: 33,122

33,122

5m: 33,122 / 1h: 0

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 59.9 tok/s 98.8 tok/s 127.3 tok/s - 86.1 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 00:46:25 ~ 00:46:35 (6.9s) 00:45:57 ~ 00:46:07 (7.0s) 00:44:46 ~ 00:44:51 (4.8s) 00:44:36 ~ 00:44:46 (5.1s) 00:46:10 ~ 00:46:20 (5.0s)

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。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) 2,492 ms 2,457 ms 2,047 ms 2,568 ms 1,995 ms

首文本(SSE delta) 2,494 ms 2,468 ms 2,049 ms - 2,032 ms

总耗时 5,740 ms 5,644 ms 3,738 ms 3,772 ms 4,105 ms

输入 10 10 10 10 33,595

输出 231 278 210 202 289

缓存创建

255

5m: 0 / 1h: 255

192

5m: 192 / 1h: 0

250

5m: 0 / 1h: 250

247

5m: 247 / 1h: 0

5m: 0 / 1h: 0

缓存读取 33,122 32,406 33,122 33,122 0

输出速度 71.2 tok/s 87.5 tok/s 124.3 tok/s 167.8 tok/s 139.4 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 00:48:51 ~ 00:48:56 (2.2s) 00:48:12 ~ 00:48:22 (7.1s) 00:48:28 ~ 00:48:33 (3.4s) 00:45:48 ~ 00:45:53 (4.5s) 00:49:27 ~ 00:49:32 (4.1s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) - 2,741 ms 1,936 ms 2,529 ms 2,343 ms

首文本(SSE delta) - 3,405 ms 1,937 ms - 2,343 ms

总耗时 - 4,705 ms 2,369 ms 2,963 ms 3,238 ms

输入 - 10 10 10 34,248

输出 - 67 63 63 69

缓存创建

5m: 61 / 1h: 0

250

5m: 0 / 1h: 250

225

5m: 225 / 1h: 0

5m: 0 / 1h: 0

缓存读取 - 32,598 33,372 33,369 0

输出速度 - 51.5 tok/s 145.8 tok/s 145.2 tok/s 77.1 tok/s

响应

失败

评测失败

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 - 00:51:14 ~ 00:51:25 (6.1s) 00:51:43 ~ 00:51:53 (5.6s) 00:49:48 ~ 00:49:58 (5.9s) 00:53:29 ~ 00:53:39 (6.8s)

提示词

查看本步 Prompt

中文回答：你叫什么、属于哪个产品、谁开发？50 字内。

HTTP 头(TTFB) - 2,982 ms 2,582 ms 2,087 ms 2,306 ms

首文本(SSE delta) - 2,983 ms 2,584 ms - 2,307 ms

总耗时 - 4,740 ms 4,565 ms 3,923 ms 5,967 ms

输入 - 10 10 10 34,670

输出 - 306 217 140 877

缓存创建

5m: 41 / 1h: 0

101

5m: 0 / 1h: 101

101

5m: 101 / 1h: 0

5m: 0 / 1h: 0

缓存读取 - 32,659 33,622 33,594 0

输出速度 - 174.2 tok/s 109.5 tok/s 76.3 tok/s 239.6 tok/s

响应

未执行

我是Claude Haiku 4.5，由Anthropic公司开发，属于Claude AI助手产品系列。

Claude Haiku 4.5 是由 Anthropic 公司打造的 AI 助手。Haiku 是 Claude 系列中最轻量的模型，专为快速、高效的对话和任务处理而设计。

我是Claude Haiku 4.5，由Anthropic公司打造的AI助手。我是Claude系列中较小的模型，设计用于快速响应各类任务，包括编程、分析和创意写作等。

我是Claude，由Anthropic开发的AI助手。协助用户进行编码、写作、分析、研究等专业任务。具备代码理解和任务规划能力。

步骤 5: knowledge_recall text

时间 - 00:52:38 ~ 00:52:58 (16.8s) 00:53:43 ~ 00:54:03 (16.3s) 00:51:25 ~ 00:51:40 (14.8s) 00:55:55 ~ 00:56:10 (11.0s)

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) - 2,010 ms 1,723 ms 1,360 ms 1,535 ms

首文本(SSE delta) - 2,011 ms 1,724 ms - 1,702 ms

总耗时 - 15,574 ms 15,319 ms 13,731 ms 9,858 ms

输入 - 10 10 10 10,885

输出 - 1,894 2,295 2,096 1,460

缓存创建

8,062

5m: 8,062 / 1h: 0

8,558

5m: 0 / 1h: 8,558

8,453

5m: 0 / 1h: 8,453

5m: 0 / 1h: 0

缓存读取 - 0 0 0 0

输出速度 - 139.6 tok/s 168.8 tok/s 169.4 tok/s 179.0 tok/s

响应

未执行

```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":"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 - 00:56:12 ~ 00:56:57 (42.7s) 00:56:49 ~ 00:57:19 (28.9s) 00:54:09 ~ 00:55:19 (68.3s) 00:59:52 ~ 01:00:02 (8.8s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=784317551835535074394268141650886709361945085658962354214944818325635479

HTTP 头(TTFB) - 1,314 ms 1,467 ms 3,189 ms 1,412 ms

首文本(SSE delta) - 1,316 ms 1,470 ms - 1,413 ms

总耗时 - 41,294 ms 27,826 ms 66,475 ms 7,996 ms

输入 - 193 193 193 209

输出 - 3,376 5,278 13,167 1,258

缓存创建

5m: 0 / 1h: 0

缓存读取 - 0 0 0 0

输出速度 - 84.4 tok/s 200.3 tok/s 208.1 tok/s 191.1 tok/s

响应

未执行

```json
{"n":72,"c":[4,5,6]}
```

```json
{"n":72,"c":[4,5,6]}
```

```json
{"n":72,"c":[4,5,6]}
```

```json
{"n":72,"c":[4,5,6]}
```

总计

总输入(含缓存)

66,519

仅前 3 步

138,668

142,640

142,476

146,626

总输出

292

仅前 3 步

5,972

8,119

15,721

4,020

总缓存创建

33,377

5m: 0 / 1h: 33,377

仅前 3 步

40,762

5m: 40,762 / 1h: 0

42,281

5m: 0 / 1h: 42,281

42,148

5m: 33,695 / 1h: 8,453

5m: 0 / 1h: 0

总缓存读取

33,122

仅前 3 步

97,663

100,116

100,085

输出速度

68.5 tok/s

仅前 3 步

99.1 tok/s

182.5 tok/s

198.0 tok/s

181.5 tok/s

总执行时间

13.5s

仅前 3 步

1m 24.5s

1m 3.1s

1m 44.3s

40.1s

总墙钟时间

5m 42.4s

含步骤间等待

仅前 3 步

13m 45.1s

含步骤间等待

14m 3.0s

含步骤间等待

12m 9.2s

含步骤间等待

16m 54.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0465

Anthropic: Claude Haiku 4.5 (部分)

$0.0717

Anthropic: Claude Haiku 4.5

$0.1037

Anthropic: Claude Haiku 4.5

$0.1257

Anthropic: Claude Haiku 4.5

$0.1667

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 identity digit_count digit_count digit_count digit_count

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

ID 格式 msg_01BiLsu8y4MtSmKYsYfZaVX4 msg_01JNiu4p9Qwt9uG9zohJn1VG msg_013EJZVPP1B4RLf6nz9HrwCf msg_01S2QsW78WrUiJbtLTYaN6sZ msg_01zO3yCL8cgK1bDbMLjnN8s0

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专用	server: Nginx x-oneapi-request-id: 202606221646271591536778268d9d6ezFp2pDO	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	—
O-Bdrk	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606221656137962102028268d9d69F93NBWB	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	—
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_011CcJej29TkFKnp7K2tXHhx
O-Max	server: nginx/1.29.8 x-oneapi-request-id: 202606221654112375624988268d9d68jJcvSqR	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	—
R-ultra	cf-ray: a0fcce58fa9de041-NRT server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606221659533006519818268d9d60cUNE1V6	id_format: msg_prefix 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_fieldsupstream:anthropic-direct-from-body	a0fcce58fa9de041-NRT

协议指纹评分（机器自动）

总分 / 100 87 100 85 65

基线状态已对照基线已对照基线已对照基线已对照基线

模型匹配 — 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 5.0 /10 ×14

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

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

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

身份(结构) — 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

身份(自由) — 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7

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

Req-ID 透传 — 0.0 /10 ×4 10.0 /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

service_tier — 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

SDK 一致 — 10.0 /10 ×2 10.0 /10 ×2 10.0 /10 ×2 10.0 /10 ×2

系统提示纯净 — 5.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8

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

流式投递 — 未计入未计入未计入未计入

综合结论

与基线相似度执行失败，不评分 87 v3.24.0 基准（参考） 85 v3.24.0 65 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 4589 ms 12562 ms 9449 ms 15720 ms 5697 ms

输出速度 68.5 tok/s 99.1 tok/s 182.5 tok/s 198.0 tok/s 181.5 tok/s

完成步骤 2/6 成功 · 1 失败 · 3 跳过 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 ⓘ

$0.0465 (部分)

$0.0717

$0.1037

$0.1257

$0.1667

缓存读取占比 ⓘ 50% (部分) 70% 70% 70% 0%

揭晓

通道

服务商 duckcoding ↗

通道cc专用

服务商 ClaudeAPI ↗

通道O-Bdrk

官方基准

服务商 ByteCatCode ↗

通道O-Max

服务商 FoxCode ↗

通道R-ultra

返回列表