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

2026/06/26 运行日期

claude-sonnet-4-6 官方基准

2 通道分数 < 70 与基线相似度

已揭晓揭晓进度

步骤 / 指标

ClaudeAPI

O-Bdrk

官方基准

SAIAi

O-Max

ByteCatCode

O-Max

SSSAiCode

R-Kiro

基本信息

服务名 Claude Code 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 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 claude-cli v2.1.160

偏差标红基于官方基准

步骤 1: ping text

时间 06:19:32 ~ 06:19:57 (22.2s) 06:19:34 ~ 06:19:39 (3.5s) 06:19:30 ~ 06:19:40 (4.9s) 06:19:28 ~ 06:19:38 (7.1s) 06:19:26 ~ 06:19:31 (4.8s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 20,637 ms 2,092 ms 3,277 ms 5,318 ms 3,462 ms

首文本(SSE delta) 20,639 ms 2,094 ms 3,279 ms - 3,605 ms

总耗时 20,873 ms 2,225 ms 3,343 ms 5,438 ms 3,671 ms

输入 3 3 3 1,255 0

输出 5 5 5 2 2

缓存创建

32,318

5m: 0 / 1h: 32,318

5m: 0 / 1h: 0

32,318

5m: 0 / 1h: 32,318

32,421

缓存读取 0 32,318 0 0 0

响应

pong

pong

pong

pong

pong

步骤 2: identity text

时间 06:21:23 ~ 06:21:33 (9.2s) 06:22:22 ~ 06:22:27 (4.2s) 06:22:36 ~ 06:22:41 (3.1s) 06:21:06 ~ 06:21:16 (5.3s) 06:23:32 ~ 06:23:37 (4.4s)

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。请按下列 3 行格式精确填空作答（label 不变，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) 7,176 ms 2,845 ms 1,765 ms 3,781 ms 3,331 ms

首文本(SSE delta) 7,178 ms 2,846 ms 1,767 ms - 3,353 ms

总耗时 7,799 ms 3,180 ms 2,207 ms 4,000 ms 3,685 ms

输入 3 3 3 115 0

输出 22 22 22 16 18

缓存创建

182

5m: 0 / 1h: 182

182

5m: 0 / 1h: 182

5m: 0 / 1h: 0

1,259

缓存读取 32,318 32,318 32,500 32,505 0

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

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

步骤 3: cutoff text

时间 06:24:10 ~ 06:24:25 (11.1s) 06:25:42 ~ 06:25:47 (2.9s) 06:24:08 ~ 06:24:13 (3.3s) 06:23:53 ~ 06:24:03 (5.6s) 06:26:43 ~ 06:26:48 (4.1s)

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) 8,694 ms 1,761 ms 2,087 ms 3,766 ms 2,732 ms

首文本(SSE delta) 8,805 ms 1,763 ms 2,191 ms - 2,941 ms

总耗时 9,379 ms 1,912 ms 2,266 ms 4,458 ms 3,056 ms

输入 3 3 3 13 0

输出 8 8 8 2 3

缓存创建

5m: 0 / 1h: 57

5m: 0 / 1h: 38

5m: 0 / 1h: 55

131

缓存读取 32,500 32,500 32,500 33,773 0

响应

2025-08

2025-08

2025-08

2025-08

2025-08

步骤 4: identity_free text

时间 06:27:21 ~ 06:27:31 (8.9s) 06:28:14 ~ 06:28:19 (3.8s) 06:27:34 ~ 06:27:44 (6.0s) 06:25:46 ~ 06:25:56 (5.2s) 06:29:12 ~ 06:29:22 (5.4s)

提示词

查看本步 Prompt

中文做个自我介绍：你是什么产品或助手？谁开发的？50 字内。

HTTP 头(TTFB) 6,336 ms 1,891 ms 4,313 ms 3,828 ms 3,446 ms

首文本(SSE delta) 6,785 ms 1,893 ms 4,315 ms - 3,470 ms

总耗时 7,414 ms 2,793 ms 4,985 ms 4,083 ms 4,592 ms

输入 3 3 3 13 0

输出 67 59 46 20 33

缓存创建

5m: 0 / 1h: 43

5m: 0 / 1h: 35

5m: 0 / 1h: 41

缓存读取 32,557 32,538 32,555 33,919 0

输出速度 106.5 tok/s 65.6 tok/s 68.7 tok/s - 29.4 tok/s

响应

我是 Claude，由 Anthropic 开发的 AI 助手。我能帮助你完成写作、分析、编程、问答等多种任务，致力于提供安全、有用的智能对话服务。

我是 Claude，由 Anthropic 开发的 AI 助手。我基于大型语言模型构建，旨在提供安全、有用、诚实的对话与任务协助。

我叫 Claude，是由 Anthropic 公司开发的 AI 助手，当前使用的模型为 claude-sonnet-4-6。

我是Claude，由Anthropic公司开发的大型语言模型AI助手。

我是 Kiro，一个由 Anthropic 开发的 AI 驱动开发环境。我基于 Claude 模型构建，帮助开发者完成代码编写、调试和软件工程任务。

步骤 5: knowledge_recall text

时间 06:29:50 ~ 06:30:00 (5.5s) 06:31:09 ~ 06:31:14 (4.8s) 06:30:50 ~ 06:30:55 (4.2s) 06:27:43 ~ 06:27:53 (5.8s) 06:33:06 ~ 06:33:11 (4.9s)

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) 2,300 ms 1,415 ms 1,479 ms 2,579 ms 1,896 ms

首文本(SSE delta) 2,301 ms 1,417 ms 1,480 ms - 1,925 ms

总耗时 4,203 ms 3,833 ms 3,223 ms 4,709 ms 4,010 ms

输入 3 3 3 653 0

输出 196 196 196 159 224

缓存创建

8,056

5m: 0 / 1h: 8,056

2,303

5m: 0 / 1h: 2,303

8,031

5m: 0 / 1h: 8,031

7,934

缓存读取 0 5,718 0 0 0

输出速度 103.0 tok/s 81.1 tok/s 112.4 tok/s 74.6 tok/s 107.4 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

{"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

{"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

{"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

{"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

步骤 6: digit_count text

时间 06:32:18 ~ 06:32:53 (30.0s) 06:34:25 ~ 06:34:40 (10.1s) 06:34:37 ~ 06:34:52 (11.7s) 06:30:52 ~ 06:31:12 (19.6s) 06:37:01 ~ 06:37:16 (11.1s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=274996333865500585788443623135994625271772992844412833170276874350477882

HTTP 头(TTFB) 15,629 ms 971 ms 845 ms 2,068 ms 1,995 ms

首文本(SSE delta) 15,630 ms 974 ms 845 ms - 2,060 ms

总耗时 28,599 ms 9,085 ms 10,728 ms 18,235 ms 10,400 ms

输入 164 164 164 168 0

输出 1,106 766 1,012 779 621

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0 0

输出速度 85.3 tok/s 94.4 tok/s 102.4 tok/s 48.2 tok/s 74.5 tok/s

响应

I'll count each digit position by position in:
`274996333865500585788443623135994625271772992844412833170276874350477882`

**Length:**
Grouping by 10s:
`2749963338` | `6550058578` | `8443623135` | `99

```json
{"n":72,"c":[4,10,6]}
```

**Verification:**

```
s = 274996333865500585788443623135994625271772992844412833170276874350477882
```

- **Length (n):** 72
- **0s:** positions 14,15,57,66 → **4**

**Counting the string:**

`274996333865500585788443623135994625271772992844412833170276874350477882`

**Length:** Stepping through each character position 1–72 → **n = 72**

**Digit counts:**

- **0s:

```json
{"n":72,"c":[4,10,6]}
```

Breakdown:

s = `274996333865500585788443623135994625271772992844412833170276874350477882`

- Total length: 72
- `0` appears at positions 14,15,57,66 → **4**
- `7` a

```json
{"n":72,"c":[4,10,6]}
```

总计

总输入(含缓存) 138,210 138,129 138,179 144,213 0

总输出 1,404 1,056 1,289 978 901

总缓存创建

40,656

5m: 0 / 1h: 40,656

2,558

5m: 0 / 1h: 2,558

40,445

5m: 0 / 1h: 40,445

41,799

总缓存读取 97,375 135,392 97,555 100,197 0

输出速度 88.3 tok/s 89.3 tok/s 102.0 tok/s 51.3 tok/s 76.0 tok/s

总执行时间 1m 26.9s 29.3s 33.2s 48.6s 34.6s

总墙钟时间

13m 21.0s

含步骤间等待

15m 5.9s

含步骤间等待

15m 21.9s

含步骤间等待

11m 43.8s

含步骤间等待

17m 50.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2033

Anthropic: Claude Sonnet 4.6

$0.0666

Anthropic: Claude Sonnet 4.6

$0.2008

Anthropic: Claude Sonnet 4.6

$0.1493

Anthropic: Claude Sonnet 4.6

>=$0.0135

Anthropic: Claude Sonnet 4.6 (成本不完整)

通道指纹

来源步骤 digit_count digit_count digit_count digit_count digit_count

平台 one-api (100%) direct (100%) direct (100%) one-api (100%) direct (75%)

上游 Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (100%) Anthropic 直连 (75%) Anthropic 直连 (75%)

CDN - - - - -

ID 格式 msg_017LiwSNARFhUttKBMYWeiaT msg_01TYL6zwWKuUUPQsnPm6N8GR msg_01G5vUEvixknmMU36faa2nN8 msg_0c8c201efd844d8e84a5025f886ae476 msg_6dd8bb8a6747409d8f04bc8c054662eb

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-Bdrk	x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 202606252232202903263198268d9d6Ie2wsDGq	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	hdr:x-new-api-versionhdr:x-oneapi-request-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody: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: 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_011CcQmtvmd4LezQJq9zDFJ3
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_011CcQmuojV99XCCXcyB2vop
O-Max	server: nginx/1.29.8 x-oneapi-request-id: 2026062522305427044548268d9d6pmEqgJlJ	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	hdr:x-oneapi-request-idbody:id:msg_prefixbody:usage.cache_fieldsupstream:anthropic-direct-from-body	—
R-Kiro	server: Photon-Edge x-amzn-requestid: 09d07327-d05a-4783-9238-f116b1d21954	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: true	body:id:msg_prefixbody:usage.cache_fieldsupstream:x-amzn-requestidupstream:anthropic-direct-from-body	1782427022811-dinvjc3dedt4v 09d07327-d05a-4783-9238-f116b1d21954

协议指纹评分（机器自动）

总分 / 100 95 100 100 72 51

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 0.0 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 10.0 /10 ×15 10.0 /10 ×15 0.0 /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 5.0 /10 ×14

sliding 5m 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 10.0 /10 ×13 5.0 /10 ×13

原生 msg-ID 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 0.0 /10 ×8 0.0 /10 ×8

系统提示纯净 10.0 /10 ×8 10.0 /10 ×8 10.0 /10 ×8 0.0 /10 ×8 未计入

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 0.0 /10 ×7

service_tier 10.0 /10 ×6 10.0 /10 ×6 10.0 /10 ×6 0.0 /10 ×6 0.0 /10 ×6

inference_geo 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 0.0 /10 ×5 0.0 /10 ×5

延迟基线 0.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

流式投递 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 未计入 10.0 /10 ×5

Req-ID 透传 0.0 /10 ×4 10.0 /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 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

综合结论

与基线相似度 95 v3.24.1 基准（参考） 100 v3.24.1 72 v3.24.1 51 v3.24.1

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 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 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 13045 ms 3838 ms 4459 ms 6821 ms 4902 ms

输出速度 88.3 tok/s 89.3 tok/s 102.0 tok/s 51.3 tok/s 76.0 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 ⓘ

$0.2033

$0.0666

$0.2008

$0.1493

>=$0.0135 (成本不完整)

缓存读取占比 ⓘ 70% 98% 71% 69% -

揭晓

通道

服务商 ClaudeAPI ↗

通道O-Bdrk

官方基准

服务商 SAIAi ↗

通道O-Max

服务商 ByteCatCode ↗

通道O-Max

服务商 SSSAiCode ↗

通道R-Kiro

返回列表