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。

4 通道 · 6 步骤

2026/06/21 运行日期

claude-haiku-4-5-20251001 官方基准

1 任务失败执行不完整

已揭晓揭晓进度

⚠ 1 通道不可测 · 已从主对比表移除点击展开原因

Aiberm / cc
```
评测失败
```

步骤 / 指标

官方基准

DawAPI

ClaudeCN

O-Max

OmniaKey

O-Max

基本信息

服务名 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

客户端 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 14:24:41 ~ 14:24:46 (4.3s) 14:24:35 ~ 14:24:45 (7.1s) 14:24:37 ~ 14:24:47 (8.2s) 14:24:35 ~ 14:24:50 (10.7s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,361 ms 3,140 ms 3,812 ms 4,067 ms

首文本(SSE delta) 2,363 ms 3,144 ms 4,213 ms 4,068 ms

总耗时 2,954 ms 3,384 ms 4,214 ms 7,084 ms

输入 10 10 10 10

输出 53 46 39 53

缓存创建

33,122

5m: 0 / 1h: 33,122

33,122

5m: 33,122 / 1h: 0

33,123

5m: 0 / 1h: 33,123

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 0 0

输出速度 89.7 tok/s 191.7 tok/s burst 17.6 tok/s

响应

pong

pong

pong

pong

步骤 2: identity text

时间 14:26:58 ~ 14:27:08 (5.2s) 14:26:29 ~ 14:26:39 (6.4s) 14:27:34 ~ 14:27:44 (6.4s) 14:28:13 ~ 14:28:23 (7.5s)

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) 1,786 ms 3,261 ms 3,349 ms 3,033 ms

首文本(SSE delta) 1,787 ms 3,263 ms 3,721 ms 3,034 ms

总耗时 4,206 ms 5,263 ms 5,067 ms 5,485 ms

输入 10 10 10 10

输出 227 216 205 210

缓存创建

233

5m: 0 / 1h: 233

226

5m: 226 / 1h: 0

219

5m: 0 / 1h: 219

233

5m: 0 / 1h: 233

缓存读取 33,122 33,122 33,123 33,122

输出速度 93.8 tok/s 108.0 tok/s 152.3 tok/s 85.7 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

步骤 3: cutoff text

时间 14:30:43 ~ 14:30:48 (4.5s) 14:28:11 ~ 14:28:16 (3.8s) 14:28:58 ~ 14:29:08 (6.7s) 14:30:13 ~ 14:30:23 (5.0s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 2,497 ms 2,679 ms 4,867 ms 3,444 ms

首文本(SSE delta) 2,499 ms 2,680 ms 4,868 ms 3,445 ms

总耗时 3,562 ms 3,033 ms 5,503 ms 3,825 ms

输入 10 10 10 10

输出 61 60 82 60

缓存创建

250

5m: 0 / 1h: 250

239

5m: 239 / 1h: 0

247

5m: 0 / 1h: 247

233

5m: 0 / 1h: 233

缓存读取 33,355 33,348 33,342 33,355

输出速度 57.4 tok/s 170.0 tok/s 129.1 tok/s 157.9 tok/s

响应

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 14:32:29 ~ 14:32:40 (7.8s) 14:31:00 ~ 14:31:15 (12.4s) 14:30:52 ~ 14:31:08 (10.8s) 14:31:48 ~ 14:31:58 (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) 2,712 ms 2,670 ms 3,466 ms 4,337 ms

首文本(SSE delta) 2,714 ms 2,671 ms 3,467 ms 4,338 ms

总耗时 6,613 ms 11,460 ms 8,816 ms 7,267 ms

输入 10 10 10 10

输出 491 1,350 727 398

缓存创建

103

5m: 0 / 1h: 103

102

5m: 102 / 1h: 0

116

5m: 0 / 1h: 116

5m: 0 / 1h: 94

缓存读取 33,605 33,587 33,589 33,588

输出速度 125.9 tok/s 153.6 tok/s 135.9 tok/s 135.9 tok/s

响应

我是Claude Haiku，由Anthropic开发的AI助手。我可以帮助编写代码、调试程序、回答技术问题，擅长软件工程领域的工作。

我是Claude，由Anthropic公司开发的AI助手。我协助编写代码、分析问题和进行对话等任务。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手，可以帮助用户进行代码编写、文件编辑、研究和其他软件工程相关的任务。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我可以帮助软件工程任务、代码编辑、问题解答和各种技术工作。

步骤 5: knowledge_recall text

时间 14:34:34 ~ 14:35:34 (57.8s) 14:35:06 ~ 14:35:26 (18.1s) 14:33:45 ~ 14:34:05 (18.3s) 14:33:29 ~ 14:33:49 (17.6s)

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) 1,517 ms 1,890 ms 2,961 ms 3,509 ms

首文本(SSE delta) 1,518 ms 1,891 ms 2,962 ms 3,510 ms

总耗时 56,796 ms 16,017 ms 14,794 ms 16,592 ms

输入 10 10 10 10

输出 5,600 2,282 1,798 2,247

缓存创建

8,817

5m: 0 / 1h: 8,817

9,657

5m: 0 / 1h: 9,657

9,050

5m: 0 / 1h: 9,050

8,698

5m: 0 / 1h: 8,698

缓存读取 0 0 0 0

输出速度 101.3 tok/s 161.5 tok/s 152.0 tok/s 171.8 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_2024":"Argentina","euro2024":"Spain

步骤 6: digit_count text

时间 14:36:52 ~ 14:37:27 (30.8s) 14:36:54 ~ 14:37:04 (5.3s) 14:36:06 ~ 14:36:36 (28.2s) 14:35:59 ~ 14:36:44 (44.2s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=598418172671649794115951305875804819948741987644190729790077903517127321

HTTP 头(TTFB) 1,382 ms 2,838 ms 2,386 ms 2,679 ms

首文本(SSE delta) 1,383 ms 2,968 ms 2,387 ms 2,680 ms

总耗时 29,881 ms 4,323 ms 24,483 ms 41,314 ms

输入 193 204 193 193

输出 3,248 135 4,653 4,204

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0

输出速度 114.0 tok/s 99.6 tok/s 210.6 tok/s 108.8 tok/s

响应

```json
{"n":72,"c":[6,12,11]}
```

I'll count the digits exactly in the string s.

Let me count each digit:

s = "598418172671649794115951305875804819948741987644190729790077903517127321"

First, let me get the total length and count t

```json
{
  "n": 72,
  "c": [6, 12, 11]
}
```

```json
{"n":72,"c":[6,12,11]}
```

总计

总输入(含缓存) 142,850 143,657 143,052 142,688

总输出 9,680 4,089 7,504 7,172

总缓存创建

42,525

5m: 0 / 1h: 42,525

43,346

5m: 33,689 / 1h: 9,657

42,755

5m: 0 / 1h: 42,755

42,380

5m: 0 / 1h: 42,380

总缓存读取 100,082 100,057 100,054 100,065

输出速度 105.5 tok/s 152.2 tok/s 180.9 tok/s 118.6 tok/s

总执行时间 1m 50.5s 53.1s 1m 18.5s 1m 33.4s

总墙钟时间

12m 46.4s

含步骤间等待

12m 29.3s

含步骤间等待

11m 59.3s

含步骤间等待

12m 9.7s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1118

Anthropic: Claude Haiku 4.5

$0.0691

Anthropic: Claude Haiku 4.5

$0.1012

Anthropic: Claude Haiku 4.5

$0.0991

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count digit_count digit_count

平台 direct (100%) one-api (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%)

CDN - - cloudflare cloudflare

ID 格式 msg_01R5d9x2x3meiCue8e646tDo msg_01NoH3YGSz1QYE9syRfFc5Z4 msg_01CTJvXoDGcFKiWnGtyYi1mF msg_013JTuqhLRWUSV6bJ3WdVDWP

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
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_011CcFwe8HdXYc1K9xJs3D4Z
cc	server: cdn x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606210636555790494958268d9d6DYYOYo0u	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-new-api-versionhdr:x-oneapi-request-idbody:id:msg_prefixbody:model:date_suffixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	—
O-Max	cf-ray: a0f0ff537d8c6686-AMS server: cloudflare x-oneapi-request-id: 20260621143610747113447VZDS3ZTf	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	a0f0ff537d8c6686-AMS
O-Max	cf-ray: a0f0ff24c84c1c0a-AMS server: cloudflare x-oneapi-request-id: 202606210636027551089748268d9d6iZXsXblr	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	a0f0ff24c84c1c0a-AMS

协议指纹评分（机器自动）

总分 / 100 99 85 98 97

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 10.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 0.0 /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 10.0 /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 9.9 /10 ×5 10.0 /10 ×5

Req-ID 透传 10.0 /10 ×4 0.0 /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

与基线相似度基准（参考） 85 v3.24.0 98 v3.24.0 97 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 17335 ms 7247 ms 10480 ms 13595 ms

输出速度 105.5 tok/s 152.2 tok/s 180.9 tok/s 118.6 tok/s

完成步骤 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 ⓘ

$0.1118

$0.0691

$0.1012

$0.0991

缓存读取占比 ⓘ 70% 70% 70% 70%

揭晓

通道

官方基准

服务商 DawAPI ↗

通道cc

服务商 ClaudeCN ↗

通道O-Max

服务商 OmniaKey ↗

通道O-Max

返回列表