能力坐标

模型不靠声量，
靠坐标说话。

每日读取官方评测，把模型能力拆成综合智能、编程与智能体三条清晰坐标。

当前坐标 / AGENTIC

智能体能力指数

衡量模型规划、工具调用与完成多步骤自主任务的能力。

2026-09-27

名次 / 模型智能体能力指数 · HIGHER IS BETTER

01Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)Anthropic · 2026-09-01
57.9
02Claude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)Anthropic · 2026-09-01
57.2
03Claude Opus 5 (Adaptive Reasoning, Max Effort)Anthropic · 2026-07-24
56.5
04Qwen3.8 Max (0902)Alibaba · 2026-09-02
56.0
05Claude Opus 5 (Adaptive Reasoning, Xhigh Effort)Anthropic · 2026-07-24
55.6
06Muse Spark 1.3 (max)Meta · 2026-09-02
55.5
07Qwen3.8-Flash-NextAlibaba · 2026-08-26
53.6
08Claude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)Anthropic · 2026-09-01
53.1
09GLM-5.3 (max)Z AI · 2026-08-18
53.1
10Grok 4.6 (high)SpaceXAI · 2026-08-12
53.0
11Grok 4.6 (xhigh)SpaceXAI · 2026-08-12
52.5
12Claude Opus 5 (Adaptive Reasoning, High Effort)Anthropic · 2026-07-24
52.3
13Muse Spark 1.3 (xhigh)Meta · 2026-09-02
51.5
14GPT-6 Astra (max)OpenAI · 2026-09-03
51.0
15GLM 5.3 FlashZ AI · 2026-08-26
50.9
16Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)Anthropic · 2026-06-09
50.7
17Grok 4.6 (medium)SpaceXAI · 2026-08-12
50.6
18Claude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)Anthropic · 2026-09-01
50.2
19GPT-5.6 Sol (max)OpenAI · 2026-07-09
50.2
20GPT-6 Astra (xhigh)OpenAI · 2026-09-03
50.2