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

2026/06/24 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

ClaudeCN

O-Max

DawAPI

0-0

O-Max

luminai

M-Kiro/Antigravity

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 09:59:15 ~ 09:59:26 (5.5s) 09:59:17 ~ 09:59:27 (5.5s) 09:59:15 ~ 09:59:26 (7.4s) 09:59:19 ~ 09:59:29 (6.2s) 09:59:21 ~ 09:59:26 (4.0s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,328 ms 3,761 ms 4,604 ms 4,646 ms 2,144 ms

首文本(SSE delta) 3,330 ms 3,762 ms 4,606 ms 4,710 ms 2,146 ms

总耗时 3,799 ms 4,202 ms 5,093 ms 4,766 ms 2,917 ms

输入 10 10 10 1,443 10

输出 43 45 50 1 47

缓存创建

33,123

5m: 0 / 1h: 33,123

33,122

5m: 33,122 / 1h: 0

33,192

5m: 0 / 1h: 33,192

4,111

5m: 4,111 / 1h: 0

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 0 23,302 0

输出速度 91.7 tok/s 102.3 tok/s 102.7 tok/s - 61.0 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 10:01:49 ~ 10:01:59 (8.2s) 10:00:55 ~ 10:01:05 (7.0s) 10:01:41 ~ 10:01:51 (8.7s) 10:03:06 ~ 10:03:16 (6.2s) 10:01:53 ~ 10:02:03 (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。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) 3,651 ms 3,304 ms 3,878 ms 4,690 ms 1,720 ms

首文本(SSE delta) 3,652 ms 3,305 ms 3,880 ms 4,757 ms 1,722 ms

总耗时 6,573 ms 6,222 ms 6,760 ms 5,092 ms 5,157 ms

输入 10 10 10 1,452 10

输出 175 256 281 28 192

缓存创建

223

5m: 0 / 1h: 223

225

5m: 225 / 1h: 0

178

5m: 0 / 1h: 178

4,136

5m: 4,136 / 1h: 0

227

5m: 0 / 1h: 227

缓存读取 33,123 33,122 33,192 23,441 33,122

输出速度 59.9 tok/s 87.8 tok/s 97.6 tok/s - 55.9 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 10:05:35 ~ 10:05:45 (5.3s) 10:04:56 ~ 10:05:01 (4.7s) 10:04:44 ~ 10:04:54 (8.3s) 10:05:08 ~ 10:05:18 (5.8s) 10:05:47 ~ 10:05:52 (3.9s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 3,364 ms 3,026 ms 6,228 ms 4,543 ms 1,807 ms

首文本(SSE delta) 3,803 ms 3,027 ms 6,229 ms 4,595 ms 1,809 ms

总耗时 4,049 ms 3,961 ms 6,860 ms 4,650 ms 2,842 ms

输入 10 10 10 1,455 10

输出 64 66 84 4 71

缓存创建

198

5m: 0 / 1h: 198

296

5m: 296 / 1h: 0

5m: 0 / 1h: 42

4,145

5m: 4,145 / 1h: 0

232

5m: 0 / 1h: 232

缓存读取 33,346 33,347 33,370 23,492 33,349

输出速度 - 70.7 tok/s 133.1 tok/s - 68.7 tok/s

响应

2025-02

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 10:07:29 ~ 10:07:39 (6.6s) 10:07:42 ~ 10:07:52 (6.5s) 10:06:08 ~ 10:06:23 (11.9s) 10:08:24 ~ 10:08:34 (6.1s) 10:08:30 ~ 10:08:40 (5.2s)

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) 3,269 ms 3,759 ms 4,295 ms 4,741 ms 1,861 ms

首文本(SSE delta) 3,270 ms 3,760 ms 4,296 ms 5,018 ms 1,864 ms

总耗时 5,390 ms 5,622 ms 10,370 ms 5,076 ms 4,170 ms

输入 10 10 10 1,457 10

输出 318 201 809 32 137

缓存创建

104

5m: 0 / 1h: 104

100

5m: 100 / 1h: 0

5m: 0 / 1h: 43

4,151

5m: 4,151 / 1h: 0

111

5m: 0 / 1h: 111

缓存读取 33,544 33,643 33,412 23,527 33,581

输出速度 150.0 tok/s 107.9 tok/s 133.2 tok/s - 59.4 tok/s

响应

我是Claude Haiku 4.5，属于Anthropic公司的Claude产品系列。由Anthropic公司开发。

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。我可以帮助编程、分析、回答问题和进行各种文本任务。

我是Anthropic公司开发的AI助手Claude。可以帮助用户编程、分析、写作等任务，提供安全、诚实、有帮助的回应。

我叫Claude Haiku 4.5，属于Claude产品家族，由Anthropic公司开发。

我是Claude，由Anthropic公司开发的AI助手。属于Claude产品系列，是一个大语言模型。

步骤 5: knowledge_recall text

时间 10:09:52 ~ 10:10:12 (16.7s) 10:10:58 ~ 10:11:48 (45.3s) 10:08:07 ~ 10:08:22 (12.1s) 10:11:56 ~ 10:12:06 (5.3s) 10:12:08 ~ 10:13:08 (58.9s)

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) 2,726 ms 3,166 ms 3,114 ms 2,931 ms 1,474 ms

首文本(SSE delta) 2,727 ms 3,167 ms 3,115 ms 2,932 ms 1,476 ms

总耗时 15,497 ms 44,489 ms 10,579 ms 4,174 ms 57,962 ms

输入 10 10 10 374 10

输出 2,185 6,033 1,266 205 3,811

缓存创建

8,584

5m: 0 / 1h: 8,584

8,562

5m: 0 / 1h: 8,562

8,051

5m: 0 / 1h: 8,051

1,064

5m: 1,064 / 1h: 0

8,447

5m: 0 / 1h: 8,447

缓存读取 0 0 0 6,033 0

输出速度 171.1 tok/s 146.0 tok/s 169.6 tok/s 165.1 tok/s 67.5 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

```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 10:13:44 ~ 10:14:04 (16.8s) 10:14:14 ~ 10:14:29 (12.8s) 10:11:12 ~ 10:12:42 (88.4s) 10:13:08 ~ 10:13:13 (4.0s) 10:14:37 ~ 10:14:57 (16.7s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=347459856118460070689730116878715342001870291376188439231217810672582456

HTTP 头(TTFB) 2,438 ms 1,617 ms 24,557 ms 2,484 ms 1,894 ms

首文本(SSE delta) 2,439 ms 1,618 ms - 2,507 ms 1,895 ms

总耗时 15,650 ms 12,035 ms 24,782 ms 2,764 ms 15,738 ms

输入 193 193 193 190 193

输出 2,172 1,915 4,683 19 2,269

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 - 0

输出速度 164.4 tok/s 183.8 tok/s - - 163.9 tok/s

响应

```json
{"n":72,"c":[8,9,4]}
```

```json
{"n":72,"c":[8,9,4]}
```

```json
{"n": 72, "c": [8, 9, 4]}
```

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

```json
{"n":72,"c":[8,9,4]}
```

总计

总输入(含缓存) 142,488 142,660 141,723 123,773 142,434

总输出 4,957 8,516 7,173 289 6,527

总缓存创建

42,232

5m: 0 / 1h: 42,232

42,305

5m: 33,743 / 1h: 8,562

41,506

5m: 0 / 1h: 41,506

17,607

5m: 17,607 / 1h: 0

42,139

5m: 0 / 1h: 42,139

总缓存读取 100,013 100,112 99,974 99,795 100,052

输出速度 155.4 tok/s 147.1 tok/s 142.0 tok/s 165.1 tok/s 83.8 tok/s

总执行时间 59.1s 1m 21.8s 2m 16.8s 33.5s 1m 34.8s

总墙钟时间

14m 48.8s

含步骤间等待

15m 12.2s

含步骤间等待

13m 26.9s

含步骤间等待

13m 54.2s

含步骤间等待

15m 36.5s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0878

Anthropic: Claude Haiku 4.5

$0.0899

Anthropic: Claude Haiku 4.5

$0.0980

Anthropic: Claude Haiku 4.5

$0.0316

Anthropic: Claude Haiku 4.5

$0.0956

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 one-api (100%) one-api (100%) direct (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_012cStRUNRKsYpHUYPhPHmcM msg_01NWRryy9aTQiNT7syUyALG1 msg_01DQj7rbaYKutBArZKn6nK1c msg_01PpkYLP8qvTWzVjHrz3K3LpGh msg_01MfAQzGYum6WLCntuaN1fyZ

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
O-Max	cf-ray: a108370d994099d3-AMS server: cloudflare x-oneapi-request-id: 20260624101345783541664yTRtJWac	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	a108370d994099d3-AMS
cc	server: cdn x-new-api-version: v1.0.0-rc.14 x-oneapi-request-id: 202606240214151938501728268d9d60QkkVfKt	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	via: 1.1 Caddy server: uvicorn, nginx/1.24.0 (Ubuntu)	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:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	ec3dc4781557a8c0c833dc17e07b37b7
M-Kiro/Antigravity	cf-ray: a108362ecc125392-AMS server: cloudflare x-oneapi-request-id: 202606240213096702026468268d9d6QMelfqrY	id_format: msg_prefix extra_fields: ["container"] has_service_tier: true has_inference_geo: false 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:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a108362ecc125392-AMS
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_011CcMH52Kdx22n89PLrQHRB

协议指纹评分（机器自动）

总分 / 100 96 90 96 72 100

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 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 0.0 /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 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 10.0 /10 ×8

系统提示纯净 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 0.0 /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 10.0 /10 ×6 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 0.0 /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 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 96 v3.24.0 90 v3.24.0 96 v3.24.0 72 v3.24.0 基准（参考）

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 8493 ms 12755 ms 10741 ms 4420 ms 14798 ms

输出速度 155.4 tok/s 147.1 tok/s 142.0 tok/s 165.1 tok/s 83.8 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 ⓘ

$0.0878

$0.0899

$0.0980

$0.0316

$0.0956

缓存读取占比 ⓘ 70% 70% 71% 81% 70%

揭晓

通道

服务商 ClaudeCN ↗

通道O-Max

服务商 DawAPI ↗

通道cc

服务商 0-0 ↗

通道O-Max

服务商 luminai ↗

通道M-Kiro/Antigravity

官方基准

返回列表