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 另含 2 任务失败

已揭晓揭晓进度

步骤 / 指标

duckcoding

cc专用

官方基准

AiTokensFlux

O-Pro

DawAPI

ByteCatCode

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 03:33:00 ~ 03:33:10 (5.0s) 03:33:06 ~ 03:33:11 (4.6s) 03:33:04 ~ 03:33:14 (5.5s) 03:33:02 ~ 03:33:12 (8.5s) 03:32:58 ~ 03:33:08 (5.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,422 ms 2,466 ms 3,601 ms 7,172 ms 3,749 ms

首文本(SSE delta) 3,424 ms 2,468 ms 3,602 ms 7,294 ms -

总耗时 3,857 ms 3,353 ms 4,635 ms 7,374 ms 3,998 ms

输入 10 10 10 10 10

输出 53 52 52 46 54

缓存创建

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,261

5m: 0 / 1h: 33,261

33,170

5m: 0 / 1h: 33,170

33,168

5m: 0 / 1h: 33,168

缓存读取 0 0 0 0 0

输出速度 122.4 tok/s 58.8 tok/s 50.3 tok/s - 216.9 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 03:35:17 ~ 03:35:27 (5.2s) 03:37:01 ~ 03:37:11 (5.1s) 03:34:15 ~ 03:34:25 (6.5s) 03:34:44 ~ 03:34:54 (7.0s) 03:34:09 ~ 03:34:19 (7.1s)

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。请只输出以下 3 行内容（label 小写，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) 2,724 ms 2,264 ms 2,234 ms 4,063 ms 3,859 ms

首文本(SSE delta) 2,725 ms 2,266 ms 2,235 ms 6,066 ms -

总耗时 4,446 ms 4,032 ms 5,298 ms 6,204 ms 6,012 ms

输入 10 10 10 10 10

输出 187 214 215 203 223

缓存创建

233

5m: 0 / 1h: 233

232

5m: 0 / 1h: 232

232

5m: 0 / 1h: 232

5m: 0 / 1h: 0

178

5m: 0 / 1h: 178

缓存读取 33,122 33,122 33,261 33,349 33,168

输出速度 108.7 tok/s 121.2 tok/s 70.2 tok/s - 103.6 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 03:38:08 ~ 03:38:13 (4.7s) 03:40:50 ~ 03:40:55 (4.3s) 03:37:25 ~ 03:47:26 (600.1s) 03:38:11 ~ 03:38:21 (6.3s) 03:37:03 ~ 03:37:13 (6.7s)

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) 2,879 ms 2,674 ms - 4,944 ms 4,726 ms

首文本(SSE delta) 2,882 ms 2,676 ms - 5,147 ms -

总耗时 3,696 ms 3,170 ms - 5,200 ms 5,156 ms

输入 10 10 - 10 10

输出 100 70 - 62 69

缓存创建

229

5m: 0 / 1h: 229

254

5m: 0 / 1h: 254

5m: 0 / 1h: 0

237

5m: 0 / 1h: 237

缓存读取 33,355 33,354 - 33,388 33,170

输出速度 122.9 tok/s 141.7 tok/s - - 160.5 tok/s

响应

2025-02

2025-02

失败

探测超时

2025-02

2025-02

步骤 4: identity_free text

时间 03:41:31 ~ 03:41:41 (5.5s) 03:44:02 ~ 03:44:12 (5.7s) - 03:40:02 ~ 03:40:12 (8.3s) 03:41:00 ~ 03:41:10 (8.4s)

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) 2,584 ms 1,997 ms - 5,543 ms 4,296 ms

首文本(SSE delta) 2,585 ms 1,998 ms - 7,219 ms -

总耗时 4,565 ms 4,435 ms - 7,288 ms 7,116 ms

输入 10 10 - 10 10

输出 281 438 - 449 515

缓存创建

142

5m: 0 / 1h: 142

104

5m: 0 / 1h: 104

33,433

5m: 0 / 1h: 33,433

600

5m: 0 / 1h: 600

缓存读取 33,584 33,608 - 0 33,170

输出速度 141.9 tok/s 179.7 tok/s - burst 182.6 tok/s

响应

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，包括代码编写、bug修复、代码审查和技术问题解答。

我是Claude Haiku 4.5，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、进行研究、代码审查等工作。

失败

评测执行异常

我是Claude Haiku，由Anthropic公司开发的AI助手。我可以帮助您完成编程任务、回答问题、进行分析和创意写作等工作。

我是 Claude Haiku 4.5，由 Anthropic 公司开发的 AI 助手。我能进行各种对话和任务，包括代码编写、分析、研究等。

步骤 5: knowledge_recall text

时间 03:42:52 ~ 03:43:07 (11.1s) 03:46:07 ~ 03:46:32 (23.6s) - 03:42:41 ~ 03:42:56 (12.8s) 03:43:43 ~ 03:44:13 (25.1s)

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) 2,566 ms 1,834 ms - 3,974 ms 2,774 ms

首文本(SSE delta) 2,567 ms 1,836 ms - 11,209 ms -

总耗时 10,192 ms 22,453 ms - 11,937 ms 23,884 ms

输入 10 10 - 10 10

输出 1,330 3,405 - 1,660 3,705

缓存创建

8,625

5m: 0 / 1h: 8,625

8,768

5m: 0 / 1h: 8,768

33,199

5m: 0 / 1h: 33,199

2,808

5m: 0 / 1h: 2,808

缓存读取 0 0 - 0 30,403

输出速度 174.4 tok/s 165.2 tok/s - burst 175.5 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

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

```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 03:45:24 ~ 03:45:44 (19.1s) 03:47:51 ~ 03:48:11 (18.0s) - 03:46:19 ~ 03:46:59 (39.0s) 03:48:09 ~ 03:48:39 (26.2s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=173488604636990978666058542751373556921712189648599309241062351789455365

HTTP 头(TTFB) 1,030 ms 847 ms - 23,347 ms 3,795 ms

首文本(SSE delta) 1,046 ms 850 ms - 35,508 ms -

总耗时 17,834 ms 17,007 ms - 38,107 ms 25,126 ms

输入 193 193 - 10 10

输出 3,551 3,061 - 3,250 3,773

缓存创建

5m: 0 / 1h: 0

623

5m: 0 / 1h: 623

31,338

5m: 0 / 1h: 31,338

缓存读取 0 0 - 30,403 0

输出速度 211.5 tok/s 189.5 tok/s - burst 176.9 tok/s

响应

```json
{"n":72,"c":[5,6,9]}
```

```json
{"n":72,"c":[5,6,9]}
```

未执行

Looking at the string: `173488604636990978666058542751373556921712189648599309241062351789455365`

Let me count each digit type:

**Length:** 72 characters

**Count of 0s:** 173488**0**463699**0**9786

I'll count the digits in the string systematically.

s = 173488604636990978666058542751373556921712189648599309241062351789455365

**Length (n):** 72 characters

**Counting 0s, 7s, 9s:**

- **0s:** 8,

总计

总输入(含缓存)

142,655

142,807

66,774

仅前 4 步

197,625

198,300

总输出

5,502

7,240

267

仅前 4 步

5,670

8,339

总缓存创建

42,351

5m: 0 / 1h: 42,351

42,480

5m: 0 / 1h: 42,480

33,493

5m: 0 / 1h: 33,493

仅前 4 步

100,425

5m: 0 / 1h: 100,425

68,329

5m: 0 / 1h: 68,329

总缓存读取

100,061

100,084

33,261

仅前 4 步

97,140

129,911

输出速度

187.4 tok/s

170.9 tok/s

65.2 tok/s

仅前 4 步

burst

173.4 tok/s

总执行时间

50.6s

1m 1.2s

10m 12.1s

仅前 4 步

1m 21.9s

1m 18.8s

总墙钟时间

12m 43.9s

含步骤间等待

15m 5.0s

含步骤间等待

14m 21.4s

含步骤间等待

仅前 4 步

13m 57.0s

含步骤间等待

15m 40.8s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0907

Anthropic: Claude Haiku 4.5

$0.0996

Anthropic: Claude Haiku 4.5

$0.0465

Anthropic: Claude Haiku 4.5 (部分)

$0.1637

Anthropic: Claude Haiku 4.5

$0.1402

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count identity digit_count digit_count

平台 one-api (100%) direct (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 - - - - -

ID 格式 msg_01JbqaPzn6WtCFXwiFam5DSg msg_01KDcdAmURkXYjq4KMg2CV1a msg_01JNbM7GsGvgWzZDhHbMjy3C msg_01BJk14BuoD5T2m5nixScSBw msg_01VqzhpCB41jD2D1dDSWtaFH

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: 202606241945259519932108268d9d6qv2MHFza	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	—
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_011CcNfPJDvNj8dWhB7YXu3n
O-Pro	server: nginx	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	—
cc	server: cdn x-new-api-version: v1.0.0-rc.14 x-oneapi-request-id: 202606241946206056333298268d9d6m8iqEk8G	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/1.29.8 x-oneapi-request-id: 202606241948106590883708268d9d6HqGeJShI	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	—

协议指纹评分（机器自动）

总分 / 100 96 98 81 94

基线状态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 8.4 /10 ×20 8.4 /10 ×20 — 0.0 /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

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 10.0 /10 ×14

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

系统提示纯净 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

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

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

延迟基线 10.0 /10 ×5 10.0 /10 ×5 — 8.2 /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

stop_reason 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

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 96 v3.24.1 基准（参考）执行失败，不评分 81 v3.24.1 94 v3.24.1

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 7432 ms 9075 ms 4967 ms 12685 ms 11882 ms

输出速度 187.4 tok/s 170.9 tok/s 65.2 tok/s burst 173.4 tok/s

完成步骤 6/6 成功 6/6 成功 2/6 成功 · 2 失败 · 2 跳过 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 ⓘ

$0.0907

$0.0996

$0.0465 (部分)

$0.1637

$0.1402

缓存读取占比 ⓘ 70% 70% 50% (部分) 49% 66%

揭晓

通道

服务商 duckcoding ↗

通道cc专用

官方基准

服务商 AiTokensFlux ↗

通道O-Pro

服务商 DawAPI ↗

通道cc

服务商 ByteCatCode ↗

通道O-Max

返回列表