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。

3 通道 · 6 步骤

2026/07/02 运行日期

claude-opus-4-8 官方基准

2 任务失败执行不完整

已揭晓揭晓进度

⚠ 2 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
```
InfAI / R-Kiro
```
评测失败
```

步骤 / 指标

AiTokensFlux

O-Pro

OmniaKey

O-Max

官方基准

基本信息

服务名 Claude Code Claude Code Claude Code

模型 claude-opus-4-8 claude-opus-4-8 claude-opus-4-8

客户端 claude-cli v2.1.195 claude-cli v2.1.195 claude-cli v2.1.195

偏差标红基于官方基准

步骤 1: ping text

时间 15:30:13 ~ 15:30:18 (3.3s) 15:30:11 ~ 15:30:16 (4.8s) 15:30:15 ~ 15:30:20 (3.4s)

提示词

查看本步 Prompt

ping. Reply with the single word: pong

HTTP 头(TTFB) 2,372 ms 2,835 ms 2,235 ms

首文本(SSE delta) 2,387 ms 2,836 ms 2,236 ms

总耗时 2,514 ms 2,979 ms 2,415 ms

输入 2,326 2,326 2,326

输出 4 4 4

缓存创建

14,449

5m: 0 / 1h: 14,449

1,711

5m: 0 / 1h: 1,711

5m: 0 / 1h: 0

缓存读取 0 12,732 14,443

响应

pong

pong

pong

步骤 2: identity text

时间 15:31:20 ~ 15:31:25 (4.4s) 15:33:00 ~ 15:33:05 (4.6s) 15:33:15 ~ 15:33:20 (3.2s)

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。严格按以下 3 行格式输出，全部小写英文，不要任何其他文字、解释、标点：
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) 3,234 ms 3,170 ms 1,951 ms

首文本(SSE delta) 3,237 ms 3,171 ms 1,952 ms

总耗时 3,601 ms 3,528 ms 2,261 ms

输入 2 2 2

输出 28 28 28

缓存创建

2,515

5m: 0 / 1h: 2,515

2,515

5m: 0 / 1h: 2,515

2,515

5m: 0 / 1h: 2,515

缓存读取 14,449 14,443 14,443

响应

vendor: anthropic
brand: claude
model: claude-opus-4-8

vendor: anthropic
brand: claude
model: claude-opus-4-8

vendor: anthropic
brand: claude
model: claude-opus-4-8

步骤 3: cutoff text

时间 15:35:06 ~ 15:35:11 (4.8s) 15:35:26 ~ 15:35:31 (4.4s) 15:35:48 ~ 15:35:53 (3.0s)

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) 3,786 ms 2,847 ms 1,917 ms

首文本(SSE delta) 3,809 ms 2,848 ms 1,919 ms

总耗时 4,053 ms 3,322 ms 2,025 ms

输入 2 2 2

输出 6 6 6

缓存创建

5m: 0 / 1h: 62

5m: 0 / 1h: 54

5m: 0 / 1h: 64

缓存读取 16,964 16,958 16,958

响应

2026-01

2026-01

2026-01

步骤 4: identity_free text

时间 15:38:07 ~ 15:38:12 (4.4s) 15:38:47 ~ 15:38:57 (9.0s) 15:37:10 ~ 15:37:15 (3.8s)

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) 2,410 ms 5,833 ms 2,144 ms

首文本(SSE delta) 2,411 ms 5,834 ms 2,145 ms

总耗时 3,629 ms 7,080 ms 2,747 ms

输入 2 2 2

输出 67 68 46

缓存创建

17,065

5m: 0 / 1h: 17,065

5m: 0 / 1h: 35

5m: 0 / 1h: 31

缓存读取 0 17,012 17,022

输出速度 55.0 tok/s 54.6 tok/s 76.4 tok/s

响应

我是 Claude，由 Anthropic 开发的 AI 助手。我能协助编程、写作、分析和解答各类问题，致力于提供有帮助、诚实且安全的回应。

我是 Claude，一款由 Anthropic 打造的 AI 助手，当前模型为 Opus 4.8，通过 Claude Agent SDK 提供智能对话与编程协助。

我是 Claude，一个由 Anthropic 开发的人工智能助手，当前模型为 Claude Opus 4.8。

步骤 5: knowledge_recall text

时间 15:39:46 ~ 15:39:56 (8.4s) 15:41:29 ~ 15:41:44 (11.1s) 15:40:10 ~ 15:40:26 (10.3s)

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) 2,040 ms 2,914 ms 2,128 ms

首文本(SSE delta) 4,485 ms 4,394 ms 3,100 ms

总耗时 7,705 ms 9,077 ms 9,280 ms

输入 2 2 2

输出 671 698 768

缓存创建

5,763

5m: 0 / 1h: 5,763

5,752

5m: 0 / 1h: 5,752

3,860

5m: 0 / 1h: 3,860

缓存读取 0 0 1,876

输出速度 burst 149.0 tok/s 124.3 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

步骤 6: digit_count text

时间 15:40:59 ~ 15:41:04 (2.8s) 15:44:55 ~ 15:45:00 (4.7s) 15:42:52 ~ 15:42:57 (2.9s)

提示词

查看本步 Prompt

Count digits exactly in s. Return {"n":len(s),"c":[count0,count7,count9]}. s=124097932860117370081536323779009443905004878184624512987159288021471552

HTTP 头(TTFB) 1,674 ms 2,569 ms 1,632 ms

首文本(SSE delta) 1,675 ms 2,570 ms 1,634 ms

总耗时 2,087 ms 2,663 ms 1,992 ms

输入 211 211 211

输出 19 19 19

缓存创建

5m: 0 / 1h: 0

缓存读取 0 0 0

响应

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

{"n":72,"c":[9,7,6]}

{"n":72,"c":[9,6,7]}

总计

总输入(含缓存) 73,812 73,757 73,757

总输出 795 823 871

总缓存创建

39,854

5m: 0 / 1h: 39,854

10,067

5m: 0 / 1h: 10,067

6,470

5m: 0 / 1h: 6,470

总缓存读取 31,413 61,145 64,742

输出速度 55.0 tok/s 129.2 tok/s 120.0 tok/s

总执行时间 28.2s 38.5s 26.7s

总墙钟时间

10m 51.0s

含步骤间等待

14m 48.9s

含步骤间等待

12m 41.9s

含步骤间等待

按官方价目重估

同等内容若直发 Anthropic 的估算 ⓘ

$0.2974

Anthropic: Claude Opus 4.8

$0.1268

Anthropic: Claude Opus 4.8

$0.1073

Anthropic: Claude Opus 4.8

通道指纹

来源步骤 digit_count digit_count digit_count

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

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

CDN - cloudflare -

ID 格式 msg_0166GggpztxczJ6NxPvWdvwR msg_01DeWEfexjAduTpHYBmGCqvr msg_01QHKqi6yfKWN9R2gu6W7VX9

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-Pro

server: nginx

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

—

O-Max

cf-ray: a14c07363ba29709-AMS
server: cloudflare
x-oneapi-request-id: 202607020744578171537508268d9d6guFj3zGi

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

a14c07363ba29709-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_011Cccra6NRhS6a5iycQ5Jnj

协议指纹评分（机器自动）

总分 / 100 85 98 100

基线状态已对照基线已对照基线已对照基线

缓存命中比 5.0 /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

延迟基线 9.6 /10 ×15 9.5 /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 10.0 /10 ×14 10.0 /10 ×14 10.0 /10 ×14

sliding 5m 7.7 /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

系统提示纯净 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

身份(自由) 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

知识截止 10.0 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

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

流式投递 0.6 /10 ×5 10.0 /10 ×5 10.0 /10 ×5

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

Req-ID 透传 0.0 /10 ×2 0.0 /10 ×2 10.0 /10 ×2

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 85 v3.26.0 98 v3.26.0 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 流式投递 buffer_dump_match
6 自由身份 identity_free_clean
7 停止原因 stop_reason_present
8 系统提示纯净 system_prompt_clean
9 Service Tier service_tier_present
10 缓存命中比 cache_hit_ratio_match
11 缓存 TTL 一致 cache_ttl_consistency
12 推理地区 inference_geo_present
13 Thinking 体量 thinking_volume_match
14 缓存连续 cache_continuity_intra
15 知识召回 knowledge_recall_match
16 延迟基线 latency_baseline_match
17 原生 msg-ID anthropic_msg_id_format
18 Sliding 缓存 cache_sliding_correctness
19 结构化身份 identity_structured_match
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 3932 ms 4775 ms 3453 ms

输出速度 55.0 tok/s 129.2 tok/s 120.0 tok/s

完成步骤 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 ⓘ

$0.2974

$0.1268

$0.1073

缓存读取占比 ⓘ 43% 83% 88%

揭晓

通道

服务商 AiTokensFlux ↗

通道O-Pro

服务商 OmniaKey ↗

通道O-Max

官方基准

返回列表