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

2026/06/18 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

LingxiCode

M-

ByteCatCode

O-Max

官方基准

AIMZ

M-Max

AnyAIGC

M-

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 19:14:56 ~ 19:15:01 (4.5s) 19:14:52 ~ 19:15:02 (6.8s) 19:15:02 ~ 19:15:07 (4.8s) 19:14:54 ~ 19:14:59 (4.9s) 19:14:58 ~ 19:15:08 (5.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,798 ms 4,068 ms 3,267 ms 2,284 ms 2,903 ms

首文本(SSE delta) 2,799 ms - 3,269 ms 2,653 ms 2,910 ms

总耗时 3,317 ms 4,202 ms 3,659 ms 2,754 ms 3,683 ms

输入 10 10 10 10 10

输出 58 41 63 56 61

缓存创建

33,124

5m: 0 / 1h: 33,124

33,170

5m: 0 / 1h: 33,170

33,122

5m: 0 / 1h: 33,122

32,785

5m: 0 / 1h: 32,785

33,261

5m: 0 / 1h: 33,261

缓存读取 0 0 0 0 0

输出速度 112.0 tok/s - 161.5 tok/s - 78.9 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 19:17:36 ~ 19:17:46 (5.5s) 19:18:58 ~ 19:19:08 (6.7s) 19:19:02 ~ 19:19:12 (5.0s) 19:17:04 ~ 19:17:25 (15.7s) 19:17:17 ~ 19:17:27 (6.8s)

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) 2,377 ms 3,910 ms 1,669 ms 2,591 ms 3,512 ms

首文本(SSE delta) 2,378 ms - 1,671 ms 10,543 ms 3,513 ms

总耗时 4,419 ms 5,381 ms 3,914 ms 11,601 ms 5,226 ms

输入 10 10 10 10 10

输出 196 180 273 754 239

缓存创建

238

5m: 0 / 1h: 238

178

5m: 0 / 1h: 178

243

5m: 0 / 1h: 243

236

5m: 0 / 1h: 236

241

5m: 0 / 1h: 241

缓存读取 33,124 33,170 33,122 32,785 33,261

输出速度 96.0 tok/s 122.4 tok/s 121.7 tok/s burst 139.5 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-3-5-sonnet

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

步骤 3: cutoff text

时间 19:19:00 ~ 19:19:05 (4.4s) 19:22:45 ~ 19:22:55 (6.3s) 19:22:34 ~ 19:22:39 (3.2s) 19:20:08 ~ 19:20:18 (5.8s) 19:18:35 ~ 19:18:45 (5.6s)

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) 2,818 ms 4,959 ms 1,671 ms 3,946 ms 2,880 ms

首文本(SSE delta) 2,819 ms - 1,673 ms 4,515 ms 3,211 ms

总耗时 3,393 ms 4,960 ms 2,186 ms 4,539 ms 3,779 ms

输入 10 10 10 10 10

输出 72 78 71 104 65

缓存创建

238

5m: 0 / 1h: 238

5m: 0 / 1h: 59

296

5m: 0 / 1h: 296

796

5m: 0 / 1h: 796

262

5m: 0 / 1h: 262

缓存读取 33,362 33,348 33,365 33,021 33,502

输出速度 125.4 tok/s burst 138.4 tok/s burst 114.4 tok/s

响应

2025-02

2025-02

2025-02

2024-04

2025-02

步骤 4: identity_free text

时间 19:21:40 ~ 19:21:50 (9.7s) 19:24:52 ~ 19:25:02 (7.1s) 19:25:01 ~ 19:25:11 (9.9s) 19:23:16 ~ 19:23:26 (7.9s) 19:22:02 ~ 19:22:22 (15.0s)

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) 2,983 ms 5,104 ms 1,919 ms 3,058 ms 7,306 ms

首文本(SSE delta) 2,984 ms - 1,921 ms 4,091 ms 7,307 ms

总耗时 8,772 ms 6,004 ms 8,872 ms 4,556 ms 13,711 ms

输入 10 10 10 10 10

输出 742 184 972 218 1,103

缓存创建

114

5m: 0 / 1h: 114

5m: 0 / 1h: 39

105

5m: 0 / 1h: 105

144

5m: 0 / 1h: 144

103

5m: 0 / 1h: 103

缓存读取 33,600 33,407 33,661 33,817 33,764

输出速度 128.2 tok/s 204.4 tok/s 139.8 tok/s - 172.2 tok/s

响应

我是Claude Code，由Anthropic开发的AI助手。专门帮助用户完成软件工程任务，包括编码、调试、代码审查、测试等工作。

Claude是由Anthropic公司开发的AI助手。我是Claude家族中的Haiku 4.5模型版本，知识截止于2025年2月。

我是Claude，Anthropic开发的AI助手。当前使用Haiku 4.5版本，知识截止2025年2月。

我叫Claude，由Anthropic公司开发。我是Claude产品系列中的AI助手，专门设计用于帮助用户完成软件工程、编程和各类问题解决任务。

我是由Anthropic公司开发的AI助手Claude Haiku 4.5。是一个智能对话模型，能在自然语言理解、编程、推理等方面提供帮助。

步骤 5: knowledge_recall text

时间 19:25:34 ~ 19:25:59 (20.2s) 19:26:18 ~ 19:26:33 (14.7s) 19:28:50 ~ 19:29:05 (11.5s) 19:26:02 ~ 19:26:12 (8.2s) 19:23:48 ~ 19:24:23 (34.2s)

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) 2,221 ms 3,089 ms 2,149 ms 1,523 ms 2,002 ms

首文本(SSE delta) 2,232 ms - 2,150 ms 6,783 ms 2,012 ms

总耗时 18,963 ms 13,396 ms 10,466 ms 7,103 ms 30,615 ms

输入 10 10 10 10 10

输出 2,648 1,949 1,572 930 4,505

缓存创建

9,074

5m: 0 / 1h: 9,074

39,548

5m: 0 / 1h: 39,548

9,356

5m: 0 / 1h: 9,356

8,797

5m: 0 / 1h: 8,797

9,449

5m: 0 / 1h: 9,449

缓存读取 0 0 0 0 0

输出速度 158.3 tok/s 189.1 tok/s 189.0 tok/s burst 157.5 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

{"facts":{"super_bowl_lviii":"Kansas City Chiefs","oscars2024_best_picture":"Oppenheimer","ucl_2024":"UNKNOWN","nba_2024":"UNKNOWN","copa_america_2024":"UNKNOWN","euro2024":"UNKNOWN","nobel_peace_2024

```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 19:28:30 ~ 19:28:35 (2.2s) 19:28:27 ~ 19:28:57 (27.0s) 19:30:39 ~ 19:31:59 (75.9s) 19:30:09 ~ 19:30:24 (13.5s) 19:27:40 ~ 19:28:10 (27.3s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=331590727907835820723056385037946791718293776123935462834620213651836894

HTTP 头(TTFB) - 3,416 ms 978 ms 1,757 ms 1,057 ms

首文本(SSE delta) - - 980 ms 1,758 ms 1,058 ms

总耗时 - 25,100 ms 74,978 ms 12,415 ms 24,961 ms

输入 - 10 193 139 193

输出 - 1,957 6,421 1,519 5,038

缓存创建

37,086

5m: 0 / 1h: 37,086

5m: 0 / 1h: 0

缓存读取 - 0 0 0 0

输出速度 - 90.3 tok/s 86.8 tok/s 142.5 tok/s 210.8 tok/s

响应

失败

评测失败

I'll count the digits in the string exactly.

String: `331590727907835820723056385037946791718293776123935462834620213651836894`

Counting:
- **Total length (n)**: 72 characters
- **Count of 0**: 6 oc

```json
{"n":72,"c":[6,9,7]}
```

```json
{"n": 72, "c": [6, 9, 7]}
```

```json
{"n":72,"c":[6,9,7]}
```

总计

总输入(含缓存)

142,924

仅前 6 步

210,065

143,513

142,570

144,086

总输出

3,716

仅前 6 步

4,389

9,372

3,581

11,011

总缓存创建

42,788

5m: 0 / 1h: 42,788

仅前 6 步

110,080

5m: 0 / 1h: 110,080

43,122

5m: 0 / 1h: 43,122

42,758

5m: 0 / 1h: 42,758

43,316

5m: 0 / 1h: 43,316

总缓存读取

100,086

仅前 6 步

99,925

100,148

99,623

100,527

输出速度

144.9 tok/s

仅前 6 步

124.3 tok/s

101.4 tok/s

142.5 tok/s

177.7 tok/s

总执行时间

46.6s

仅前 6 步

1m 8.6s

1m 50.4s

56.0s

1m 34.1s

总墙钟时间

13m 38.8s

含步骤间等待

仅前 6 步

14m 5.0s

含步骤间等待

16m 57.3s

含步骤间等待

15m 30.0s

含步骤间等待

13m 12.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0821

Anthropic: Claude Haiku 4.5 (部分)

$0.1696

Anthropic: Claude Haiku 4.5

$0.1110

Anthropic: Claude Haiku 4.5

$0.0815

Anthropic: Claude Haiku 4.5

$0.1195

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 knowledge_recall 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_012VHnAQuB6tzZ88vLa82vfh msg_01VZYqBm1Wp2TmUjCwanygV3 msg_01RyDEvPXmAiCWtAtB2G5B6f msg_01M82DufMQEMgzBoeWjLvfwb msg_01HibvLgw7nWLagxutfzspj8

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
M-	cf-ray: a0d9ef298ae0e041-NRT server: cloudflare set-cookie: server_name_session=6268b44f10666c8b08965a0a61ec74a7; Max-Age=86400; httponly; path=/ x-oneapi-request-id: 20260618112536431611432cirwlt9N	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	a0d9ef298ae0e041-NRT
O-Max	server: nginx/1.29.8 x-oneapi-request-id: 202606181128296051990918268d9d6LNdRp11I	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_011CcAecmJ3pJX3LefHHygZ2
M-Max	cf-ray: a0d9f5e26e09be6e-AMS server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606181130109703259688268d9d6RKtzWpYU	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-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	a0d9f5e26e09be6e-AMS
M-	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	—

协议指纹评分（机器自动）

总分 / 100 94 97 77 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

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

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

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

身份(结构) — 10.0 /10 ×7 10.0 /10 ×7 6.7 /10 ×7 10.0 /10 ×7

知识截止 — 10.0 /10 ×7 10.0 /10 ×7 0.0 /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 — 10.0 /10 ×8 10.0 /10 ×8 5.0 /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 94 v3.24.0 基准（参考） 77 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 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 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 7773 ms 9841 ms 17346 ms 7161 ms 13663 ms

输出速度 144.9 tok/s 124.3 tok/s 101.4 tok/s 142.5 tok/s 177.7 tok/s

完成步骤 5/6 成功 · 1 失败 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 ⓘ

$0.0821 (部分)

$0.1696

$0.1110

$0.0815

$0.1195

缓存读取占比 ⓘ 70% (部分) 48% 70% 70% 70%

揭晓

通道

服务商 LingxiCode ↗

通道M-

服务商 ByteCatCode ↗

通道O-Max

官方基准

服务商 AIMZ ↗

通道M-Max

服务商 AnyAIGC ↗

通道M-

返回列表