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

2026/06/18 运行日期

claude-sonnet-4-6 官方基准

1 通道分数 < 70 另含 2 任务失败

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

WorldBase.ai

M-

官方基准

ClaudeCN

O-Max

Novaris

O-Max

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 13:08:58 ~ 13:09:08 (5.6s) 13:09:07 ~ 13:09:17 (7.6s) 13:08:56 ~ 13:09:06 (5.7s) 13:09:00 ~ 13:09:05 (4.0s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,901 ms 6,430 ms 4,053 ms 2,832 ms

首文本(SSE delta) 4,050 ms 6,445 ms 4,054 ms 2,833 ms

总耗时 4,053 ms 6,725 ms 4,242 ms 3,037 ms

输入 33,707 3 3 3

输出 2 5 5 5

缓存创建

32,318

5m: 0 / 1h: 32,318

32,319

5m: 0 / 1h: 32,319

32,318

5m: 0 / 1h: 32,318

缓存读取 - 0 0 0

响应

pong

pong

pong

pong

步骤 2: identity text

时间 13:11:34 ~ 13:11:39 (4.1s) 13:13:03 ~ 13:13:23 (4.1s) 13:13:01 ~ 13:13:11 (6.2s) 13:12:34 ~ 13:12:39 (3.2s)

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) 2,781 ms 2,119 ms 3,468 ms -

首文本(SSE delta) 3,082 ms 2,120 ms 3,469 ms -

总耗时 3,083 ms 2,725 ms 4,920 ms -

输入 33,916 3 3 -

输出 16 22 22 -

缓存创建

189

5m: 0 / 1h: 189

189

5m: 0 / 1h: 189

缓存读取 - 32,318 32,319 -

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

失败

评测失败

步骤 3: cutoff text

时间 13:14:14 ~ 13:14:24 (5.0s) 13:17:21 ~ 13:17:26 (3.6s) 13:17:07 ~ 13:17:17 (5.1s) -

提示词

查看本步 Prompt

你掌握的世界知识截止于何时？YYYY-MM 格式作答，不解释。

HTTP 头(TTFB) 3,306 ms 2,415 ms 3,407 ms -

首文本(SSE delta) 3,407 ms 2,506 ms 3,408 ms -

总耗时 3,409 ms 2,598 ms 3,701 ms -

输入 33,974 3 3 -

输出 3 8 8 -

缓存创建

5m: 0 / 1h: 55

缓存读取 - 32,507 32,508 -

响应

2025-08

2025-08

2025-08

未执行

步骤 4: identity_free text

时间 13:16:29 ~ 13:16:39 (6.9s) 13:19:45 ~ 13:19:50 (4.8s) 13:19:42 ~ 13:19:52 (7.2s) -

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) 3,306 ms 2,787 ms 3,365 ms -

首文本(SSE delta) 3,460 ms 2,789 ms 3,365 ms -

总耗时 4,595 ms 3,701 ms 5,091 ms -

输入 34,017 3 3 -

输出 49 44 65 -

缓存创建

5m: 0 / 1h: 39

5m: 0 / 1h: 43

缓存读取 - 32,562 32,563 -

输出速度 43.2 tok/s 48.2 tok/s 37.7 tok/s -

响应

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

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。

我是 Claude，由 Anthropic 开发的 AI 助手。我基于 claude-sonnet-4-6 模型，能够帮助你完成写作、编程、分析、问答等各类任务。

未执行

步骤 5: knowledge_recall text

时间 13:19:22 ~ 13:19:32 (6.4s) 13:22:13 ~ 13:22:18 (4.4s) 13:23:25 ~ 13:23:40 (12.5s) -

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) 2,110 ms 1,501 ms 11,396 ms -

首文本(SSE delta) 2,187 ms 1,503 ms 11,397 ms -

总耗时 4,142 ms 3,384 ms 11,403 ms -

输入 8,705 3 3 -

输出 157 205 205 -

缓存创建

8,034

5m: 0 / 1h: 8,034

8,059

5m: 0 / 1h: 8,059

缓存读取 - 0 0 -

输出速度 80.3 tok/s 109.0 tok/s burst -

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

未执行

步骤 6: digit_count text

时间 13:21:13 ~ 13:21:29 (13.6s) 13:26:06 ~ 13:26:21 (12.9s) 13:24:49 ~ 13:25:04 (13.6s) -

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=555642786766858590606009162858921765039914665569300126217824788151192870

HTTP 头(TTFB) 1,923 ms 1,244 ms 2,359 ms -

首文本(SSE delta) 2,015 ms 1,245 ms 2,408 ms -

总耗时 12,580 ms 11,924 ms 12,498 ms -

输入 237 164 164 -

输出 493 915 777 -

缓存创建

5m: 0 / 1h: 0

缓存读取 - 0 0 -

输出速度 46.7 tok/s 85.7 tok/s 77.0 tok/s -

响应

```json
{"n":72,"c":[8,6,7]}
```

## Counting Process

**String:** `555642786766858590606009162858921765039914665569300126217824788151192870`

### Length
Counting all characters: **n = 72**

### Digit-by-digit scan (positions for 0, 7

```json
{"n":72,"c":[8,6,7]}
```

未执行

总计

总输入(含缓存)

144,556

138,201

138,234

32,321

仅前 2 步

总输出

720

1,199

1,082

仅前 2 步

总缓存创建

40,635

5m: 0 / 1h: 40,635

40,665

5m: 0 / 1h: 40,665

32,318

5m: 0 / 1h: 32,318

仅前 2 步

总缓存读取

97,387

97,390

仅前 2 步

输出速度

51.2 tok/s

86.4 tok/s

71.3 tok/s

仅前 2 步

总执行时间

41.7s

37.3s

50.3s

7.2s

仅前 2 步

总墙钟时间

12m 30.4s

含步骤间等待

17m 14.1s

含步骤间等待

16m 7.7s

含步骤间等待

3m 39.4s

含步骤间等待

仅前 2 步

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.4445

Anthropic: Claude Sonnet 4.6

$0.2001

Anthropic: Claude Sonnet 4.6

$0.1985

Anthropic: Claude Sonnet 4.6

$0.1213

Anthropic: Claude Sonnet 4.6 (部分)

通道指纹

来源步骤 digit_count digit_count digit_count ping

平台 direct (50%) direct (100%) one-api (100%) one-api (100%)

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

CDN cloudflare - cloudflare -

ID 格式 msg_118b00b6ec0c434a9fb4867c92d5527b msg_01J6NgtfoTca8XF6v6C9ouuH msg_01WstviqMRb9Yvfredz6HSvX msg_01U5qmxgUU9nCmurKEqaZPVY

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: a0d7d9759faf66ab-AMS server: cloudflare	id_format: msg_prefix has_service_tier: false has_inference_geo: false model_has_date_suffix: false usage_has_cache_fields: false	hdr:server:cloudflarehdr:cf-raybody:id:msg_prefix	c70bb249-7dea-43d0-832c-edb94f9239fb a0d7d9759faf66ab-AMS
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_011CcAApRbCfYt66xCQ9xHHw
O-Max	cf-ray: a0d7deb60c2afe96-AMS server: cloudflare x-oneapi-request-id: 20260618132450801734784hfoRNxnT	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-oneapi-request-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a0d7deb60c2afe96-AMS
O-Max	server: nginx x-oneapi-request-id: 202606180509016458977458268d9d6AZAOIER2	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-oneapi-request-idbody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	—

协议指纹评分（机器自动）

总分 / 100 48 100 95

基线状态已对照基线已对照基线已对照基线

缓存命中比 0.0 /10 ×20 10.0 /10 ×20 10.0 /10 ×20 —

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

sliding 5m 5.0 /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 —

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

身份(结构) 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 —

身份(自由) 0.0 /10 ×7 10.0 /10 ×7 10.0 /10 ×7 —

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

inference_geo 0.0 /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 —

流式投递 10.0 /10 ×5 10.0 /10 ×5 0.0 /10 ×5 —

Req-ID 透传 0.0 /10 ×4 10.0 /10 ×4 0.0 /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 —

综合结论

与基线相似度 48 v3.23.0 基准（参考） 95 v3.23.0 执行失败，不评分

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 5310 ms 5176 ms 6976 ms 3037 ms

输出速度 51.2 tok/s 86.4 tok/s 71.3 tok/s -

完成步骤 6/6 成功 6/6 成功 6/6 成功 1/6 成功 · 1 失败 · 4 跳过

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.4445

$0.2001

$0.1985

$0.1213 (部分)

缓存读取占比 ⓘ 0% 70% 70% 0% (部分)

揭晓

通道

服务商 WorldBase.ai ↗

通道M-

官方基准

服务商 ClaudeCN ↗

通道O-Max

服务商 Novaris ↗

通道O-Max

返回列表