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。

3 通道 · 6 步骤

2026/06/20 运行日期

claude-haiku-4-5-20251001 官方基准

2 任务失败执行不完整

已揭晓揭晓进度

⚠ 2 通道不可测 · 已从主对比表移除点击展开原因

DBAI / M-
```
评测失败
```
LingxiCode / M-
```
评测失败
```

步骤 / 指标

SSSAiCode

O-Max

FoxCode

官方基准

基本信息

服务名 Claude Code Claude Code Claude Code

模型 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

偏差标红基于官方基准

步骤 1: ping text

时间 21:51:09 ~ 21:51:14 (2.7s) 21:51:11 ~ 21:51:16 (3.7s) 21:51:13 ~ 21:51:18 (3.4s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 1,149 ms 2,329 ms 1,935 ms

首文本(SSE delta) 1,467 ms 2,330 ms 1,937 ms

总耗时 1,634 ms 2,744 ms 2,315 ms

输入 10 10 10

输出 38 51 55

缓存创建

32,785

5m: 0 / 1h: 32,785

33,122

5m: 33,122 / 1h: 0

33,122

5m: 0 / 1h: 33,122

缓存读取 0 0 0

输出速度 227.5 tok/s 123.2 tok/s 145.5 tok/s

响应

pong

pong

pong

步骤 2: identity text

时间 21:54:41 ~ 21:54:51 (9.1s) 21:52:37 ~ 21:52:42 (4.8s) 21:53:22 ~ 21:53:32 (4.9s)

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) 1,955 ms 1,994 ms 1,677 ms

首文本(SSE delta) 6,531 ms 1,995 ms 1,679 ms

总耗时 6,729 ms 3,788 ms 3,843 ms

输入 10 10 10

输出 487 215 247

缓存创建

218

5m: 0 / 1h: 218

178

5m: 178 / 1h: 0

235

5m: 0 / 1h: 235

缓存读取 32,785 33,122 33,122

输出速度 burst 119.9 tok/s 114.1 tok/s

响应

vendor: anthropic
brand: claude
model: claude-3-5-sonnet-20241022

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 21:58:02 ~ 21:58:07 (2.8s) 21:56:01 ~ 21:56:06 (4.0s) 21:55:59 ~ 21:56:04 (3.6s)

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) 1,572 ms 2,269 ms 1,973 ms

首文本(SSE delta) 1,803 ms 2,270 ms 1,981 ms

总耗时 1,846 ms 2,948 ms 2,670 ms

输入 10 10 10

输出 53 70 81

缓存创建

529

5m: 0 / 1h: 529

5m: 42 / 1h: 0

289

5m: 0 / 1h: 289

缓存读取 33,003 33,300 33,357

输出速度 burst 103.2 tok/s 117.6 tok/s

响应

2024-04

2025-02

2025-02

步骤 4: identity_free text

时间 22:00:38 ~ 22:00:48 (9.4s) 21:59:23 ~ 21:59:33 (5.4s) 21:59:04 ~ 21:59:09 (4.4s)

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) 1,749 ms 1,987 ms 2,113 ms

首文本(SSE delta) 8,360 ms 1,992 ms 2,116 ms

总耗时 8,567 ms 4,449 ms 3,386 ms

输入 10 10 10

输出 862 410 233

缓存创建

5m: 0 / 1h: 93

5m: 43 / 1h: 0

123

5m: 0 / 1h: 123

缓存读取 33,532 33,342 33,646

输出速度 burst 166.9 tok/s 183.5 tok/s

响应

我是Claude，由Anthropic公司开发，属于Claude产品线的AI助手。

我是Claude，由Anthropic公司开发的AI助手。我可以帮助用户进行对话、编程、分析、写作等各种任务。我致力于提供有帮助、诚实和无害的协助。

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，包括代码编写、调试、测试和问题解答等工作。

步骤 5: knowledge_recall text

时间 22:04:42 ~ 22:05:12 (27.1s) 22:01:17 ~ 22:01:52 (32.9s) 22:02:04 ~ 22:02:29 (21.0s)

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) 984 ms 1,686 ms 1,425 ms

首文本(SSE delta) 25,933 ms 1,688 ms 1,427 ms

总耗时 26,125 ms 32,134 ms 20,051 ms

输入 10 10 10

输出 3,351 4,662 3,064

缓存创建

9,105

5m: 0 / 1h: 9,105

8,065

5m: 8,065 / 1h: 0

8,620

5m: 0 / 1h: 8,620

缓存读取 0 0 0

输出速度 burst 153.1 tok/s 164.5 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_20

步骤 6: digit_count text

时间 22:07:05 ~ 22:07:30 (20.1s) 22:03:10 ~ 22:03:45 (30.2s) 22:04:14 ~ 22:05:14 (55.8s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=832106116538017003987860816897803791877807051522156062211777799328814245

HTTP 头(TTFB) 909 ms 1,107 ms 1,049 ms

首文本(SSE delta) 912 ms 1,108 ms 1,050 ms

总耗时 19,182 ms 29,495 ms 54,832 ms

输入 139 193 193

输出 3,961 2,514 4,952

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0

输出速度 216.8 tok/s 88.6 tok/s 92.1 tok/s

响应

```json
{"n":72,"c":[9,11,5]}
```

```json
{"n":72,"c":[9,11,5]}
```

```json
{"n": 72, "c": [9, 11, 5]}
```

总计

总输入(含缓存) 142,239 141,457 142,757

总输出 8,752 7,922 8,632

总缓存创建

42,730

5m: 0 / 1h: 42,730

41,450

5m: 41,450 / 1h: 0

42,389

5m: 0 / 1h: 42,389

总缓存读取 99,320 99,764 100,125

输出速度 216.9 tok/s 123.4 tok/s 112.2 tok/s

总执行时间 1m 11.1s 1m 21.1s 1m 33.2s

总墙钟时间

16m 21.2s

含步骤间等待

12m 34.2s

含步骤间等待

14m 1.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1073

Anthropic: Claude Haiku 4.5

$0.0822

Anthropic: Claude Haiku 4.5

$0.1064

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count digit_count

平台 direct (100%) direct (100%) direct (100%)

上游 Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%)

CDN - cloudflare -

ID 格式 msg_019uua9iK9LXjWB1TC7FWMvk msg_01EDwhTdXVvhnyHKPdySfspB msg_016UTyyfbWejnoB3qrxtid4x

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: a0eb56817b50cba2-LAX
server: Photon-Edge
set-cookie: _cfuvid=NXkPADqj1CdrkeXsYmyeHlTRPQ52QSBEoJ_fxMkkAoA-1781964426.477076-1.0.1.1-hJjChVTMvsHWsQrCUeLDDA54cS6VqdHLroqWDrocJTk; 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

1781964426433-v11215aa9fcgj
req_011CcEeATDu2k6VvgHePJTg1
a0eb56817b50cba2-LAX

cf-ray: a0eb50c32ff4e393-NRT
server: cloudflare
set-cookie: _cfuvid=rAKoAYks9LQmWKftduP_ypfyvjhLRuIY_LRIyOM4DW8-1781964191.486906-1.0.1.1-AMXYT_WWbwtio_qHfFvYHfUa7_YwxrQW.N8FchQ.B.Y; 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_011CcEds8UJv3EUVm5PUxq5G
a0eb50c32ff4e393-NRT

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_011CcEdwrDeHbUM8Y11UaAzs

协议指纹评分（机器自动）

总分 / 100 87 91 100

基线状态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

缓存 TTL 一致 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

缓存连续性 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

原生 msg-ID 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8

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

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

知识截止 0.0 /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

service_tier 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

延迟基线 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

Req-ID 透传 10.0 /10 ×4 10.0 /10 ×4 10.0 /10 ×4

stop_reason 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3

SDK 一致 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

综合结论

与基线相似度 87 v3.24.0 91 v3.24.0 基准（参考）

雷达图例

共 21 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 档位思考量 tier_thinking_volume_match
21 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 10681 ms 12593 ms 14516 ms

输出速度 216.9 tok/s 123.4 tok/s 112.2 tok/s

完成步骤 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 ⓘ

$0.1073

$0.0822

$0.1064

缓存读取占比 ⓘ 70% 71% 70%

揭晓

通道

服务商 SSSAiCode ↗

通道O-Max

服务商 FoxCode ↗

通道cc

官方基准

返回列表