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

2026/06/19 运行日期

claude-haiku-4-5-20251001 官方基准

1 通道分数 < 70 另含 1 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

AnyRouter / cc
```
评测失败
```

步骤 / 指标

SudoCode

M-Max

Novaris

O-Max

官方基准

luminai

M-Kiro/Antigravity

基本信息

服务名 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

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 13:31:34 ~ 13:31:39 (4.5s) 13:31:38 ~ 13:31:43 (3.6s) 13:31:40 ~ 13:31:45 (4.0s) 13:31:32 ~ 13:31:42 (6.5s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,939 ms 2,378 ms 2,405 ms 4,723 ms

首文本(SSE delta) 2,944 ms 2,379 ms 2,407 ms 4,762 ms

总耗时 3,265 ms 2,695 ms 2,888 ms 4,831 ms

输入 10 10 10 1,443

输出 48 51 61 1

缓存创建

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

33,122

5m: 0 / 1h: 33,122

4,111

5m: 4,111 / 1h: 0

缓存读取 0 0 0 23,301

输出速度 149.5 tok/s 161.4 tok/s 126.8 tok/s -

响应

pong

pong

pong

pong

步骤 2: identity text

时间 13:33:09 ~ 13:33:19 (6.3s) 13:35:22 ~ 13:35:32 (5.0s) 13:35:20 ~ 13:35:25 (4.2s) 13:34:26 ~ 13:34:36 (7.3s)

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。自报身份，输出严格遵守以下 3 行格式（不允许 markdown、代码块、额外文本）：
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) 2,687 ms 1,725 ms 1,732 ms 4,566 ms

首文本(SSE delta) 2,689 ms 1,725 ms 1,733 ms 4,612 ms

总耗时 5,146 ms 4,125 ms 3,263 ms 5,009 ms

输入 10 10 10 1,453

输出 250 244 170 28

缓存创建

242

5m: 0 / 1h: 242

245

5m: 0 / 1h: 245

255

5m: 0 / 1h: 255

4,138

5m: 4,138 / 1h: 0

缓存读取 33,122 33,122 33,122 23,452

输出速度 101.8 tok/s 101.7 tok/s 111.1 tok/s -

响应

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

vendor: anthropic
brand: claude
model: claude-haiku-4-5-20251001

步骤 3: cutoff text

时间 13:36:54 ~ 13:36:59 (4.7s) 13:39:06 ~ 13:39:11 (3.4s) 13:38:52 ~ 13:38:57 (3.3s) 13:38:31 ~ 13:38:41 (5.5s)

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) 3,000 ms 2,050 ms 1,834 ms 4,062 ms

首文本(SSE delta) 3,002 ms 2,051 ms 1,835 ms 4,163 ms

总耗时 3,621 ms 2,585 ms 2,303 ms 4,231 ms

输入 10 10 10 1,456

输出 65 65 63 4

缓存创建

292

5m: 0 / 1h: 292

267

5m: 0 / 1h: 267

193

5m: 0 / 1h: 193

4,148

5m: 4,148 / 1h: 0

缓存读取 33,364 33,367 33,377 23,507

输出速度 105.0 tok/s 121.7 tok/s 134.6 tok/s -

响应

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 13:40:49 ~ 13:40:59 (5.5s) 13:40:27 ~ 13:40:37 (7.8s) 13:41:28 ~ 13:41:38 (8.8s) 13:39:58 ~ 13:40:08 (7.4s)

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) 2,538 ms 2,266 ms 2,418 ms 4,734 ms

首文本(SSE delta) 2,539 ms 2,267 ms 2,420 ms 5,316 ms

总耗时 4,323 ms 7,055 ms 7,776 ms 6,144 ms

输入 10 10 10 1,458

输出 266 667 604 77

缓存创建

107

5m: 0 / 1h: 107

107

5m: 0 / 1h: 107

103

5m: 0 / 1h: 103

4,154

5m: 4,154 / 1h: 0

缓存读取 33,656 33,634 33,570 23,545

输出速度 149.1 tok/s 139.3 tok/s 112.8 tok/s 93.0 tok/s

响应

我是Claude Code，由Anthropic公司开发的AI编程助手。我专门帮助用户进行代码编写、调试、重构和优化等软件工程任务。

我是Claude，由Anthropic公司开发的AI助手。可以帮助代码编程、文本分析、问题解答等任务。致力于提供有帮助、安全、诚实的回复。

我是Claude，由Anthropic公司开发的AI助手。Claude是Anthropic的产品系列。

我是 Claude Haiku 4.5，由 Anthropic 公司开发的 AI 助手。我是一个快速、轻量级的对话模型，可以帮助用户进行软件工程任务、代码编写、分析和其他专业工作。

步骤 5: knowledge_recall text

时间 13:44:30 ~ 13:45:00 (29.5s) 13:44:23 ~ 13:44:43 (15.6s) 13:44:37 ~ 13:44:57 (18.2s) 13:43:48 ~ 13:43:53 (4.8s)

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) 2,182 ms 1,370 ms 1,593 ms 2,675 ms

首文本(SSE delta) 2,183 ms 1,371 ms 1,594 ms 2,715 ms

总耗时 28,403 ms 14,897 ms 17,058 ms 3,842 ms

输入 10 10 10 377

输出 3,917 2,123 2,075 195

缓存创建

8,647

5m: 0 / 1h: 8,647

9,026

5m: 0 / 1h: 9,026

8,895

5m: 0 / 1h: 8,895

1,073

5m: 1,073 / 1h: 0

缓存读取 0 0 0 6,082

输出速度 149.4 tok/s 157.0 tok/s 134.2 tok/s 173.0 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

```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

步骤 6: digit_count text

时间 13:48:22 ~ 13:48:48 (20.6s) 13:48:39 ~ 13:49:10 (25.2s) 13:46:07 ~ 13:46:28 (19.3s) 13:45:14 ~ 13:45:24 (5.1s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=989816309650167970494113215937292632176818089471854229353812435304423474

HTTP 头(TTFB) 1,185 ms 2,309 ms 1,391 ms 3,121 ms

首文本(SSE delta) 1,188 ms 2,310 ms 1,392 ms 3,134 ms

总耗时 19,472 ms 24,278 ms 18,363 ms 3,562 ms

输入 193 193 193 189

输出 3,365 4,457 3,261 19

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 -

输出速度 184.0 tok/s 202.9 tok/s 192.2 tok/s -

响应

```json
{"n":74,"c":[5,6,10]}
```

```json
{"n":72,"c":[5,6,9]}
```

```json
{"n":73,"c":[5,6,9]}
```

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

总计

总输入(含缓存) 142,795 143,133 142,880 123,887

总输出 7,911 7,607 6,234 324

总缓存创建

42,410

5m: 0 / 1h: 42,410

42,767

5m: 0 / 1h: 42,767

42,568

5m: 0 / 1h: 42,568

17,624

5m: 17,624 / 1h: 0

总缓存读取 100,142 100,123 100,069 99,887

输出速度 159.2 tok/s 174.7 tok/s 154.8 tok/s 139.1 tok/s

总执行时间 1m 11.1s 1m 0.6s 57.8s 36.5s

总墙钟时间

17m 14.0s

含步骤间等待

17m 32.0s

含步骤间等待

14m 48.0s

含步骤间等待

13m 52.1s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.1028

Anthropic: Claude Haiku 4.5

$0.1017

Anthropic: Claude Haiku 4.5

$0.0946

Anthropic: Claude Haiku 4.5

$0.0318

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count digit_count digit_count

平台 one-api (100%) one-api (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_0164QVjJzwwtcvg3jVuwSMD9 msg_01An5q5YWAEvKXXeh8CVfvAQ msg_0194UbtgpzNrAdkWYhmZnJxF msg_01yHhe3t1hTTgaMnTW5AbiYnYL

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-Max	server: nginx x-new-api-version: v1.0.0-rc.11 x-oneapi-request-id: 20260619134824210653521MGXfDbFg	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	—
O-Max	server: nginx x-oneapi-request-id: 202606190548408756370768268d9d6QWbzWbyD	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-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_011CcC69uSibrDXgk8xfczo6
M-Kiro/Antigravity	cf-ray: a0e03a011c78760b-AMS server: cloudflare x-oneapi-request-id: 202606190545161763329808268d9d6Wfglo8Ss	id_format: msg_prefix extra_fields: ["container"] has_service_tier: true has_inference_geo: false model_has_date_suffix: true usage_has_cache_fields: true	hdr:x-oneapi-request-idhdr:server:cloudflarehdr:cf-raybody:id:msg_prefixbody:usage.service_tierbody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	a0e03a011c78760b-AMS

协议指纹评分（机器自动）

总分 / 100 98 97 100 70

基线状态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

缓存 TTL 一致 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

缓存连续性 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

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

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

身份(结构) 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

身份(自由) 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

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

延迟基线 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5 8.6 /10 ×5

Req-ID 透传 0.0 /10 ×4 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 10.0 /10 ×3

SDK 一致 10.0 /10 ×2 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未计入

综合结论

与基线相似度 98 v3.24.0 97 v3.24.0 基准（参考） 70 v3.24.0

雷达图例

共 23 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 自报身份一致 self_identity_consistency
20 信封自报一致 envelope_self_report_match
21 中文遵循 instruction_following_lang
22 档位思考量 tier_thinking_volume_match
23 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 10705 ms 9273 ms 8609 ms 4603 ms

输出速度 159.2 tok/s 174.7 tok/s 154.8 tok/s 139.1 tok/s

完成步骤 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 ⓘ

$0.1028

$0.1017

$0.0946

$0.0318

缓存读取占比 ⓘ 70% 70% 70% 81%

揭晓

通道

服务商 SudoCode ↗

通道M-Max

服务商 Novaris ↗

通道O-Max

官方基准

服务商 luminai ↗

通道M-Kiro/Antigravity

返回列表