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

2026/06/28 运行日期

claude-sonnet-4-6 官方基准

3 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

DBAI / M-
```
评测失败
```

步骤 / 指标

SAIAi

O-Max

0-0

O-Max

官方基准

FoxCode

R-ultra

基本信息

服务名 Claude Code Claude Code Claude Code Claude Code

模型 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6 claude-sonnet-4-6

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 15:37:42 ~ 15:37:52 (5.7s) 15:37:46 ~ 15:37:56 (9.2s) 15:37:48 ~ 15:37:53 (4.3s) 15:37:40 ~ 15:37:50 (6.9s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,754 ms 5,367 ms 3,142 ms 1,887 ms

首文本(SSE delta) 2,756 ms 5,368 ms 3,144 ms 1,893 ms

总耗时 2,836 ms 5,599 ms 3,388 ms 2,356 ms

输入 3 3 3 42

输出 5 5 5 15

缓存创建

32,318

5m: 0 / 1h: 32,318

32,507

5m: 32,507 / 1h: 0

5m: 0 / 1h: 0

32,973

5m: 32,973 / 1h: 0

缓存读取 0 0 32,318 0

响应

pong

pong

pong

pong

步骤 2: identity text

时间 15:41:07 ~ 15:41:12 (4.2s) 15:40:25 ~ 15:40:35 (7.5s) 15:39:40 ~ 15:39:45 (3.3s) 15:40:16 ~ 15:40:26 (5.2s)

提示词

查看本步 Prompt

输出 3 行 metadata 描述你自己（当前 API 模型）。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。禁止任何额外字符：
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) 1,970 ms 4,330 ms 1,850 ms 3,717 ms

首文本(SSE delta) 1,971 ms 4,331 ms 1,852 ms 3,718 ms

总耗时 3,151 ms 4,792 ms 2,314 ms 4,357 ms

输入 3 3 3 88

输出 22 22 22 72

缓存创建

5m: 0 / 1h: 0

7,691

5m: 7,691 / 1h: 0

189

5m: 0 / 1h: 189

463

5m: 463 / 1h: 0

缓存读取 32,507 25,004 32,318 32,973

输出速度 - - - 112.7 tok/s

响应

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

vendor: anthropic
brand: claude
model: claude-sonnet-4-6

步骤 3: cutoff text

时间 15:43:34 ~ 15:43:39 (2.9s) 15:44:03 ~ 15:44:13 (7.4s) 15:41:05 ~ 15:41:10 (3.4s) 15:43:44 ~ 15:43:49 (4.1s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 1,728 ms 4,697 ms 2,171 ms 2,966 ms

首文本(SSE delta) 1,888 ms 4,698 ms 2,339 ms 2,967 ms

总耗时 1,918 ms 5,575 ms 2,388 ms 3,217 ms

输入 3 3 3 43

输出 8 8 8 37

缓存创建

5m: 0 / 1h: 0

32,752

5m: 32,752 / 1h: 0

5m: 0 / 1h: 38

33,875

5m: 33,875 / 1h: 0

缓存读取 32,545 0 32,507 0

输出速度 - - - 148.0 tok/s

响应

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 15:45:27 ~ 15:45:32 (3.7s) 15:47:22 ~ 15:47:32 (6.7s) 15:43:10 ~ 15:43:15 (4.4s) 15:47:13 ~ 15:47:18 (1.9s)

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) 1,748 ms - 2,812 ms -

首文本(SSE delta) 2,148 ms - 2,814 ms -

总耗时 2,692 ms - 3,460 ms -

输入 3 - 3 -

输出 62 - 44 -

缓存创建

5m: 0 / 1h: 43

5m: 0 / 1h: 39

缓存读取 32,545 - 32,545 -

输出速度 114.0 tok/s - 68.1 tok/s -

响应

我是 Claude，由 Anthropic 开发的 AI 助手。我基于大型语言模型构建，能够帮助用户完成写作、编程、分析、问答等各类任务。

失败

评测失败

我是 Claude，由 Anthropic 打造的 AI 助手，当前运行的模型版本为 claude-sonnet-4-6。

失败

评测失败

步骤 5: knowledge_recall text

时间 15:49:17 ~ 15:49:22 (4.3s) - 15:45:47 ~ 15:45:52 (4.1s) -

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) 1,513 ms - 1,252 ms -

首文本(SSE delta) 1,514 ms - 1,253 ms -

总耗时 3,365 ms - 3,133 ms -

输入 3 - 3 -

输出 204 - 204 -

缓存创建

2,321

5m: 0 / 1h: 2,321

8,017

5m: 0 / 1h: 8,017

缓存读取 5,718 - 0 -

输出速度 110.2 tok/s - 108.5 tok/s -

响应

{"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","nobel

未执行

{"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","nobel

未执行

步骤 6: digit_count text

时间 15:51:28 ~ 15:51:43 (12.6s) - 15:47:57 ~ 15:48:12 (13.7s) -

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=370283664078043078378766330028123528700096008098988366870234678675549964

HTTP 头(TTFB) 1,056 ms - 1,093 ms -

首文本(SSE delta) 1,058 ms - 1,095 ms -

总耗时 11,646 ms - 12,798 ms -

输入 164 - 164 -

输出 1,030 - 1,261 -

缓存创建

5m: 0 / 1h: 0

缓存读取 0 - 0 -

输出速度 97.3 tok/s - 107.8 tok/s -

响应

```json
{"n":72,"c":[13,9,5]}
```

**Verification:**

`s = 370283664078043078378766330028123528700096008098988366870234678675549964`

**Length:** 72 digits

**0s** (positions 3,10,13,16,27,28,38,39,40

未执行

## Counting digits in:
`370283664078043078378766330028123528700096008098988366870234678675549964`

### Length Count (groups of 10):
```
3 7 0 2 8 3 6 6 4 0  → 10
7 8 0 4 3 0 7 8 3 7  → 20
8 7 6 6 3 3

未执行

总计

总输入(含缓存)

138,176

97,963

仅前 4 步

138,150

100,457

仅前 4 步

总输出

1,331

仅前 4 步

1,544

124

仅前 4 步

总缓存创建

34,682

5m: 0 / 1h: 34,682

72,950

5m: 72,950 / 1h: 0

仅前 4 步

8,283

5m: 0 / 1h: 8,283

67,311

5m: 67,311 / 1h: 0

仅前 4 步

总缓存读取

103,315

25,004

仅前 4 步

129,688

32,973

仅前 4 步

输出速度

99.8 tok/s

仅前 4 步

106.1 tok/s

122.6 tok/s

仅前 4 步

总执行时间

33.4s

30.8s

仅前 4 步

33.3s

18.1s

仅前 4 步

总墙钟时间

14m 0.4s

含步骤间等待

9m 45.6s

含步骤间等待

仅前 4 步

10m 24.0s

含步骤间等待

9m 37.6s

含步骤间等待

仅前 4 步

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1816

Anthropic: Claude Sonnet 4.6

$0.1790

Anthropic: Claude Sonnet 4.6 (部分)

$0.0937

Anthropic: Claude Sonnet 4.6

$0.1700

Anthropic: Claude Sonnet 4.6 (部分)

通道指纹

来源步骤 digit_count cutoff digit_count cutoff

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

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

CDN - - - cloudflare

ID 格式 msg_01WeBq8MSiWTzh74zpQbNXWm msg_01VGv9XPHxQr7knJJKzhd89j msg_01CKxGnCu8NV23ayTXLj8dSJ msg_01qmRcn1cJHZdzJO8GDNqh7E

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	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: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcVHzTtn8r61fApTRrfgc
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: false usage_has_cache_fields: true	hdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	428c29c3c1004fc8cd92cc8d4b6e6cc7
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: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	req_011CcVHix31N5cLqJiKYBhLp
R-ultra	cf-ray: a12b0ff27fc3af8b-NRT server: cloudflare x-new-api-version: v0.0.0 x-oneapi-request-id: 202606280743456410430338268d9d6cbL5CVC8	id_format: msg_prefix has_service_tier: true has_inference_geo: true model_has_date_suffix: false 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:usage.cache_fieldsupstream:anthropic-direct-from-body	a12b0ff27fc3af8b-NRT

协议指纹评分（机器自动）

总分 / 100 100 100

基线状态已对照基线已对照基线

缓存命中比 10.0 /10 ×20 — 10.0 /10 ×20 —

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

模型匹配 10.0 /10 ×14 — 10.0 /10 ×14 —

缓存连续性 10.0 /10 ×14 — 10.0 /10 ×14 —

sliding 5m 10.0 /10 ×13 — 10.0 /10 ×13 —

原生 msg-ID 10.0 /10 ×8 — 10.0 /10 ×8 —

系统提示纯净 10.0 /10 ×8 — 10.0 /10 ×8 —

身份(结构) 10.0 /10 ×7 — 10.0 /10 ×7 —

知识截止 10.0 /10 ×7 — 10.0 /10 ×7 —

身份(自由) 10.0 /10 ×7 — 10.0 /10 ×7 —

service_tier 10.0 /10 ×6 — 10.0 /10 ×6 —

inference_geo 10.0 /10 ×5 — 10.0 /10 ×5 —

延迟基线 10.0 /10 ×5 — 10.0 /10 ×5 —

流式投递 10.0 /10 ×5 — 10.0 /10 ×5 —

Req-ID 透传 10.0 /10 ×4 — 10.0 /10 ×4 —

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

SDK 一致 10.0 /10 ×2 — 10.0 /10 ×2 —

综合结论

与基线相似度 100 v3.25.0 执行失败，不评分基准（参考）执行失败，不评分

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 流式投递 buffer_dump_match
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 缓存连续 cache_continuity_intra
13 知识召回 knowledge_recall_match
14 延迟基线 latency_baseline_match
15 原生 msg-ID anthropic_msg_id_format
16 Sliding 缓存 cache_sliding_correctness
17 结构化身份 identity_structured_match
18 信封自报一致 envelope_self_report_match
19 档位思考量 tier_thinking_volume_match
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 4268 ms 5322 ms 4580 ms 3310 ms

输出速度 99.8 tok/s - 106.1 tok/s 122.6 tok/s

完成步骤 6/6 成功 3/6 成功 · 1 失败 · 2 跳过 6/6 成功 3/6 成功 · 1 失败 · 2 跳过

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1816

$0.1790 (部分)

$0.0937

$0.1700 (部分)

缓存读取占比 ⓘ 75% 26% (部分) 94% 33% (部分)

揭晓

通道

服务商 SAIAi ↗

通道O-Max

服务商 0-0 ↗

通道O-Max

官方基准

服务商 FoxCode ↗

通道R-ultra

返回列表