模型对比 | RPDiag

测试说明点击展开 5 步 quick-probe-v1 详解

同一会话连续 5 步，覆盖 5 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。步与步之间随机延迟 1–4 分钟。第 3 步（cutoff）累计跨度 > 5 分钟保住 sliding cache 检测；全流程总跨度 > 6 分钟容纳新增的第 5 步。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 · 5 步骤

2026/06/15 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

AnyAIGC

M-

SSSAiCode

O-Max-L2

SudoCode

M-Max

官方基准

FoxCode

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 18:37:35 ~ 18:37:45 (7.4s) 18:37:31 ~ 18:37:36 (4.6s) 18:37:37 ~ 18:37:42 (3.7s) 18:37:39 ~ 18:37:44 (3.5s) 18:37:33 ~ 18:37:39 (4.3s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 5,329 ms 1,373 ms 2,180 ms 2,131 ms 2,776 ms

首文本(SSE delta) 5,331 ms 1,764 ms 2,182 ms 2,132 ms 2,778 ms

总耗时 6,049 ms 1,807 ms 2,602 ms 2,421 ms 3,240 ms

输入 8 10 10 10 10

输出 2 56 53 41 47

缓存创建

7,324

5m: 0 / 1h: 7,324

32,785

5m: 0 / 1h: 32,785

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,122

5m: 33,122 / 1h: 0

缓存读取 29,386 0 0 0 0

输出速度 - burst 126.2 tok/s 141.9 tok/s 101.7 tok/s

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 18:39:35 ~ 18:39:45 (5.8s) 18:41:25 ~ 18:41:35 (9.3s) 18:38:48 ~ 18:38:53 (4.7s) 18:41:10 ~ 18:41:21 (5.1s) 18:39:11 ~ 18:39:21 (5.3s)

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。严格按以下 3 行格式输出，全部小写英文，不要任何其他文字、解释、标点：
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) 3,659 ms 2,436 ms 2,003 ms 1,953 ms 2,605 ms

首文本(SSE delta) 3,679 ms 8,392 ms 2,004 ms 1,954 ms 2,606 ms

总耗时 4,601 ms 8,613 ms 3,768 ms 4,110 ms 4,490 ms

输入 8 10 10 10 10

输出 22 571 213 233 173

缓存创建

6,807

5m: 0 / 1h: 6,807

249

5m: 0 / 1h: 249

246

5m: 0 / 1h: 246

234

5m: 0 / 1h: 234

191

5m: 191 / 1h: 0

缓存读取 28,999 32,785 33,122 33,122 33,122

输出速度 - burst 120.7 tok/s 108.1 tok/s 91.8 tok/s

响应

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

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 18:42:41 ~ 18:42:46 (4.5s) 18:43:02 ~ 18:43:07 (3.0s) 18:40:24 ~ 18:40:29 (3.3s) 18:42:34 ~ 18:42:39 (4.1s) 18:42:11 ~ 18:42:21 (5.5s)

提示词

查看本步 Prompt

What's your training data cutoff (YYYY-MM only, no prose)?

HTTP 头(TTFB) 3,310 ms 1,691 ms 1,825 ms 2,069 ms 3,740 ms

首文本(SSE delta) 3,311 ms 2,157 ms 1,826 ms 2,071 ms 3,741 ms

总耗时 3,320 ms 2,242 ms 2,378 ms 3,089 ms 4,612 ms

输入 8 10 10 10 10

输出 4 79 72 68 81

缓存创建

6,791

5m: 0 / 1h: 6,791

601

5m: 0 / 1h: 601

262

5m: 0 / 1h: 262

276

5m: 0 / 1h: 276

5m: 49 / 1h: 0

缓存读取 28,880 33,034 33,368 33,356 33,313

输出速度 - - 130.4 tok/s 66.8 tok/s 93.0 tok/s

响应

2025-02

2025-04

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 18:44:20 ~ 18:44:25 (4.9s) 18:46:58 ~ 18:47:08 (7.1s) 18:42:21 ~ 18:42:31 (9.4s) 18:45:35 ~ 18:45:40 (4.2s) 18:45:33 ~ 18:45:48 (11.8s)

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) 3,462 ms 1,708 ms 1,950 ms 2,150 ms 2,328 ms

首文本(SSE delta) 3,641 ms 6,049 ms 1,955 ms 2,152 ms 2,330 ms

总耗时 3,642 ms 6,196 ms 8,395 ms 3,113 ms 10,794 ms

输入 8 10 10 10 10

输出 57 653 882 151 1,190

缓存创建

7,049

5m: 0 / 1h: 7,049

117

5m: 0 / 1h: 117

114

5m: 0 / 1h: 114

108

5m: 0 / 1h: 108

5m: 39 / 1h: 0

缓存读取 28,675 33,635 33,630 33,632 33,362

输出速度 burst burst 137.0 tok/s 157.1 tok/s 140.6 tok/s

响应

我是Claude Code，由Anthropic开发的AI编程助手。我帮助开发者完成软件工程任务，包括编写代码、调试问题、重构代码等工作。

我是Claude，由Anthropic开发的AI助手。可以帮助编程、数据分析、文本处理等各种任务。

我是Claude，Anthropic开发的AI助手。可帮助编程、分析、写作等任务，致力于以诚实、有益、安全的方式为用户服务。

我叫Claude Haiku，属于Claude产品系列，由Anthropic公司开发。

我是Anthropic开发的Claude Haiku 4.5，AI助手，可帮助编程、文本分析和问题解答。

步骤 5: knowledge_recall text

时间 18:45:42 ~ 18:45:52 (6.0s) 18:49:18 ~ 18:49:38 (17.1s) 18:45:44 ~ 18:46:09 (23.3s) 18:47:34 ~ 18:47:54 (15.7s) 18:47:16 ~ 18:47:31 (14.4s)

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) 3,911 ms 1,298 ms 1,600 ms 1,959 ms 2,519 ms

首文本(SSE delta) 4,752 ms 16,130 ms 1,601 ms 1,961 ms 2,520 ms

总耗时 4,790 ms 16,252 ms 21,991 ms 14,683 ms 13,700 ms

输入 8 10 10 10 10

输出 169 2,736 3,421 1,968 1,918

缓存创建

2,013

5m: 0 / 1h: 2,013

9,023

5m: 0 / 1h: 9,023

9,244

5m: 0 / 1h: 9,244

8,509

5m: 0 / 1h: 8,509

8,054

5m: 8,054 / 1h: 0

缓存读取 8,490 0 0 0 0

输出速度 burst burst 167.8 tok/s 154.7 tok/s 171.6 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

总计

总输入(含缓存) 154,454 142,279 143,158 142,409 141,302

总输出 254 4,095 4,641 2,461 3,409

总缓存创建

29,984

5m: 0 / 1h: 29,984

42,775

5m: 0 / 1h: 42,775

42,988

5m: 0 / 1h: 42,988

42,249

5m: 0 / 1h: 42,249

41,455

5m: 41,455 / 1h: 0

总缓存读取 124,430 99,454 100,120 100,110 99,797

输出速度 burst burst 157.0 tok/s 143.5 tok/s 149.1 tok/s

总执行时间 28.6s 41.1s 44.4s 32.5s 41.3s

总墙钟时间

8m 16.8s

含步骤间等待

12m 6.9s

含步骤间等待

8m 31.9s

含步骤间等待

10m 14.9s

含步骤间等待

9m 57.8s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0512

Anthropic: Claude Haiku 4.5

$0.0839

Anthropic: Claude Haiku 4.5

$0.0870

Anthropic: Claude Haiku 4.5

$0.0752

Anthropic: Claude Haiku 4.5

$0.0595

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 knowledge_recall knowledge_recall knowledge_recall knowledge_recall knowledge_recall

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

ID 格式 msg_uOdGrutKVoVoQjmDqt0CUbFP msg_01GQuLjdVmujR8F55tU85wWc msg_01BFsZ2W2qLGVcvFanGLXGzR msg_018iqt1QWWN4MBxxLt6QoW5V msg_01GJ7BCuTih3mXwYjykwVW8v

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-	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	—
O-Max-L2	cf-ray: a0c101e9ba484711-LAX server: Photon-Edge set-cookie: _cfuvid=GkN_6RYZNQ04Zi3U2VX7HOUhws0P4fULEvvuHNklS7U-1781520559.637472-1.0.1.1-giZ0S1aUFAdGZlOuIDj0gx.bZnhx3PRHD.BRvoc_I5g; HttpOnly; SameSite=None; Secure; Path=/; Domain=api.anthropic.com	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	1781520559594-7ec034e4eqnob req_011Cc4v385LvrDfmsATL1NCN a0c101e9ba484711-LAX
M-Max	server: nginx x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 20260615184546256645798qQWgZr28	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_011Cc4uuVAPTRhy9wQkZ6MVp
cc	cf-ray: a0c0feee0b0e7963-NRT server: cloudflare set-cookie: _cfuvid=kx1ymfurrwA6L9MuiH2tjAnX0S1PNRwDR4WMLZ9x_9s-1781520438.1737254-1.0.1.1-VAAR5ArYMmFjFg0cqASRXJA2KsX8bhz3am6NjMVz69E; HttpOnly; SameSite=None; Secure; Path=/; Domain=api.anthropic.com	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: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	req_011Cc4utB2hQDhzR1E8iLAZU a0c0feee0b0e7963-NRT

协议指纹评分（机器自动）

总分 / 100 81 92 98 100 91

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 10.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 0.0 /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 0.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8

系统提示纯净 0.0 /10 ×8 5.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8

身份(结构) 10.0 /10 ×7 6.7 /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 10.0 /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 10.0 /10 ×4 0.0 /10 ×4 10.0 /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 — — — — —

综合结论

与基线相似度 81 v3.16.0 92 v3.16.0 98 v3.16.0 基准（参考） 91 v3.16.0

雷达图例

共 20 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 中文遵循 instruction_following_lang
20 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 4480 ms 7022 ms 7827 ms 5483 ms 7367 ms

输出速度 burst burst 157.0 tok/s 143.5 tok/s 149.1 tok/s

完成步骤 5/5 成功 5/5 成功 5/5 成功 5/5 成功 5/5 成功

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0512

$0.0839

$0.0870

$0.0752

$0.0595

缓存读取占比 ⓘ 81% 70% 70% 70% 71%

揭晓

通道

服务商 AnyAIGC ↗

通道M-

服务商 SSSAiCode ↗

通道O-Max-L2

服务商 SudoCode ↗

通道M-Max

官方基准

服务商 FoxCode ↗

通道cc

返回列表