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

2026/06/27 运行日期

claude-haiku-4-5-20251001 官方基准

1 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

Aiberm / cc
```
评测失败
```

步骤 / 指标

FastCode

0-0

O-Max

SAIAi

O-Max

官方基准

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 19:59:52 ~ 19:59:57 (4.7s) 19:59:50 ~ 20:00:05 (11.3s) 19:59:54 ~ 19:59:59 (3.2s) 19:59:56 ~ 20:00:01 (3.5s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 3,031 ms 8,221 ms 1,829 ms 1,821 ms

首文本(SSE delta) 3,033 ms 8,292 ms 1,833 ms 1,822 ms

总耗时 3,339 ms 9,420 ms 2,173 ms 2,205 ms

输入 10 10 10 10

输出 54 41 51 41

缓存创建

33,122

5m: 0 / 1h: 33,122

33,192

5m: 0 / 1h: 33,192

33,122

5m: 0 / 1h: 33,122

5m: 0 / 1h: 0

缓存读取 0 0 0 33,122

输出速度 176.5 tok/s 36.3 tok/s 150.0 tok/s 107.0 tok/s

响应

pong

pong

pong

pong

步骤 2: identity text

时间 20:02:20 ~ 20:02:30 (8.6s) 20:01:09 ~ 20:01:19 (7.9s) 20:01:20 ~ 20:01:25 (4.1s) 20:01:18 ~ 20:01:28 (5.2s)

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) 5,103 ms 4,009 ms 1,916 ms 1,896 ms

首文本(SSE delta) 5,104 ms 4,011 ms 1,918 ms 1,898 ms

总耗时 7,430 ms 6,129 ms 3,070 ms 4,273 ms

输入 10 10 10 10

输出 257 230 155 228

缓存创建

238

5m: 0 / 1h: 238

182

5m: 0 / 1h: 182

235

5m: 0 / 1h: 235

225

5m: 0 / 1h: 225

缓存读取 33,122 33,192 33,122 33,122

输出速度 110.5 tok/s 108.6 tok/s 134.5 tok/s 96.0 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 20:04:06 ~ 20:04:16 (5.2s) 20:02:26 ~ 20:02:36 (6.4s) 20:04:22 ~ 20:04:27 (3.0s) 20:03:03 ~ 20:03:08 (3.7s)

提示词

查看本步 Prompt

Knowledge cutoff date? Reply with YYYY-MM only.

HTTP 头(TTFB) 2,761 ms 3,641 ms 1,532 ms 2,000 ms

首文本(SSE delta) 2,762 ms 3,642 ms 1,534 ms 2,001 ms

总耗时 4,025 ms 4,575 ms 2,083 ms 2,707 ms

输入 10 10 10 10

输出 68 99 62 65

缓存创建

280

5m: 0 / 1h: 280

5m: 0 / 1h: 61

178

5m: 0 / 1h: 178

268

5m: 0 / 1h: 268

缓存读取 33,360 33,374 33,357 33,347

输出速度 53.8 tok/s 106.1 tok/s 112.9 tok/s 92.1 tok/s

响应

2025-02

2025-02

2025-02

2025-02

步骤 4: identity_free text

时间 20:06:06 ~ 20:06:16 (7.2s) 20:06:20 ~ 20:06:30 (8.1s) 20:08:12 ~ 20:08:17 (4.1s) 20:06:31 ~ 20:06:41 (6.0s)

提示词

查看本步 Prompt

用中文说说你的身份：什么产品？由谁打造？50 字以内。

HTTP 头(TTFB) 2,687 ms 3,312 ms 1,477 ms 1,598 ms

首文本(SSE delta) 2,688 ms 3,313 ms 1,478 ms 1,599 ms

总耗时 4,727 ms 6,572 ms 3,122 ms 5,064 ms

输入 10 10 10 10

输出 280 416 273 422

缓存创建

106

5m: 0 / 1h: 106

5m: 0 / 1h: 43

100

5m: 0 / 1h: 100

5m: 0 / 1h: 99

缓存读取 33,640 33,435 33,535 33,615

输出速度 137.3 tok/s 127.6 tok/s 166.1 tok/s 121.8 tok/s

响应

我是Claude Haiku 4.5，由Anthropic公司开发的AI助手。可以帮助用户完成软件工程任务、代码分析、问题解答等工作。

我是Claude Haiku 4.5，由Anthropic开发的AI助手。我可以帮助用户进行软件工程任务、代码编写、调试和分析等工作。

我是Claude Haiku 4.5，由Anthropic公司打造的AI助手模型。我是Claude系列中的轻量级版本，设计用于快速、高效地处理各种任务。

我是Claude，由Anthropic公司开发的AI助手。我能够进行对话、分析代码、回答问题、完成各种软件工程任务。

步骤 5: knowledge_recall text

时间 20:07:48 ~ 20:08:28 (35.6s) 20:10:06 ~ 20:10:21 (14.7s) 20:09:35 ~ 20:09:55 (17.4s) 20:09:57 ~ 20:10:17 (17.9s)

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) 2,229 ms 2,575 ms 1,319 ms 1,667 ms

首文本(SSE delta) 2,230 ms 2,577 ms 1,320 ms 1,669 ms

总耗时 33,518 ms 12,286 ms 16,456 ms 16,987 ms

输入 10 10 10 10

输出 4,154 1,206 2,336 1,702

缓存创建

8,644

5m: 0 / 1h: 8,644

8,076

5m: 0 / 1h: 8,076

8,526

5m: 0 / 1h: 8,526

3,033

5m: 0 / 1h: 3,033

缓存读取 0 0 0 5,721

输出速度 132.8 tok/s 124.2 tok/s 154.3 tok/s 111.1 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_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 20:09:54 ~ 20:10:09 (13.8s) 20:12:46 ~ 20:13:11 (23.9s) 20:13:16 ~ 20:13:36 (18.0s) 20:12:17 ~ 20:12:32 (10.5s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=674968343549325323644634674770557613412537249440740866133361296639890041

HTTP 头(TTFB) 1,180 ms 6,275 ms 887 ms 891 ms

首文本(SSE delta) 1,181 ms 6,276 ms 888 ms 893 ms

总耗时 11,733 ms 22,314 ms 16,989 ms 9,518 ms

输入 193 193 193 193

输出 1,879 3,053 3,258 1,531

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0 0

输出速度 178.1 tok/s 190.4 tok/s 202.3 tok/s 177.5 tok/s

响应

```json
{"n":72,"c":[5,7,6]}
```

```json
{"n":72,"c":[5,7,6]}
```

```json
{"n": 72, "c": [5, 7, 6]}
```

```json
{"n":72,"c":[5,7,7]}
```

总计

总输入(含缓存) 142,755 141,798 142,418 142,795

总输出 6,692 5,045 6,135 3,989

总缓存创建

42,390

5m: 0 / 1h: 42,390

41,554

5m: 0 / 1h: 41,554

42,161

5m: 0 / 1h: 42,161

3,625

5m: 0 / 1h: 3,625

总缓存读取 100,122 100,001 100,014 138,927

输出速度 140.1 tok/s 152.0 tok/s 175.7 tok/s 129.2 tok/s

总执行时间 1m 15.1s 1m 12.4s 49.9s 46.8s

总墙钟时间

10m 17.0s

含步骤间等待

13m 21.1s

含步骤间等待

13m 42.1s

含步骤间等待

12m 36.0s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.0967

Anthropic: Claude Haiku 4.5

$0.0874

Anthropic: Claude Haiku 4.5

$0.0936

Anthropic: Claude Haiku 4.5

$0.0386

Anthropic: Claude Haiku 4.5

通道指纹

来源步骤 digit_count digit_count digit_count digit_count

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

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

CDN - - - -

ID 格式 msg_01MFg5RB3pj9w6neNgDup8sc msg_01BgeQDU9MQ77nEGuRDD7oNa msg_011SP3q7SeMsXxMF98epzaAx msg_019QJobAYChwzH6TMxs3QCWE

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
cc	server: nginx	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	—
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: true usage_has_cache_fields: true	hdr:viabody:id:msg_prefixbody:usage.service_tierbody:usage.inference_geobody:model:date_suffixbody:usage.cache_fieldsbody:extra_fieldsupstream:anthropic-direct-from-body	5da4eb36bb11a65a0d7e424ebae5ac6c
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: 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_011CcTk9Vx5Kww8KoupnRc4W
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_011CcTk59yvUWX4UEQMqXDoL

协议指纹评分（机器自动）

总分 / 100 86 86 89 88

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 0.0 /10 ×20 0.0 /10 ×20 0.0 /10 ×20 0.0 /10 ×20

缓存 TTL 一致 10.0 /10 ×15 10.0 /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 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 10.0 /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 10.0 /10 ×5

延迟基线 9.4 /10 ×5 7.7 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

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

stop_reason 10.0 /10 ×3 10.0 /10 ×3 10.0 /10 ×3 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 86 v3.25.0 86 v3.25.0 89 v3.25.0 基准（参考）

雷达图例

共 22 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 档位思考量 tier_thinking_volume_match
22 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 10795 ms 10216 ms 7316 ms 6792 ms

输出速度 140.1 tok/s 152.0 tok/s 175.7 tok/s 129.2 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 ⓘ

$0.0967

$0.0874

$0.0936

$0.0386

缓存读取占比 ⓘ 70% 71% 70% 97%

揭晓

通道

服务商 FastCode ↗

通道cc

服务商 0-0 ↗

通道O-Max

服务商 SAIAi ↗

通道O-Max

官方基准

返回列表