The AI tools index that doesn't waste your time.不浪费你时间的 AI 工具索引。
Agent · weight 5%智能体 · 权重 5%

τ²-bench

official site官网 ↗

What it tests: Tool-calling agent evaluation in realistic customer-service domains (airline, retail, telecom). The agent must use APIs, follow policies, and handle multi-turn user interactions.

考什么:仿真客服场景的工具调用评测(航空/零售/电信):agent 必须正确调 API、遵守业务规则、处理多轮用户对话。

Why it counts: The most practical agent benchmark — it predicts whether a model will reliably drive tools in production, which raw chat benchmarks can't tell you.

为什么算数:最实用的 Agent benchmark——能预测模型在生产环境调工具的可靠性,这是聊天类 benchmark 看不出来的。

Maintainer维护方Sierra(Anthropic 系创业公司)
License许可MIT
Reference point (prior-gen SOTA)参考点(上一代 SOTA)70
◈

Standings实时排名

32 models with sourced scores 个模型有溯源成绩
#Model模型Score成绩Index指数Dated日期Measured by测评方
1GLM 5.2Z.ai99.1⚡max100—3rd-party第三方source ↗×4
2Claude Fable 5Anthropic98.599—3rd-party第三方source ↗×5⚠±17
3Step 3.7 FlashStepFun98.599—3rd-party第三方source ↗×3
4Grok 4.3xAI97.799—3rd-party第三方source ↗×3
5GLM 5.1Z.ai97.7⚡high99—3rd-party第三方source ↗×3
6Grok 4.20xAI96.597—3rd-party第三方source ↗
7DeepSeek V4 Pro 0423DeepSeek96.297—3rd-party第三方source ↗×3
8Gemini 3.1 Pro PreviewGoogle95.696—3rd-party第三方source ↗×5⚠±3.7
9DeepSeek V4 Flash 0423DeepSeek95.6⚡high96—3rd-party第三方source ↗×2
10Gemini 3.5 FlashGoogle95.396—3rd-party第三方source ↗×3⚠±20.9
11Qwen3.7 MaxQwen94.796—3rd-party第三方source ↗×3⚠±4.1
12Claude Opus 4.8Anthropic94.495—3rd-party第三方source ↗×4⚠±17.8
13Mistral Medium 3.5Mistral AI94.295—3rd-party第三方source ↗×3
14Seed-2.0-LiteByteDance Seed94.295—vendor-reported厂商自报source ↗
15Kimi K3Moonshot AI9394—3rd-party第三方source ↗×2⚠±22.1
16Qwen3.7 PlusQwen9394—3rd-party第三方source ↗
17Kimi K2 ThinkingMoonshot AI93⚡high94—3rd-party第三方source ↗
18DeepSeek V3.2DeepSeek91⚡high922025-12-013rd-party第三方source ↗×3⚠±5.2
19Kimi K2.7 CodeMoonshot AI90.191—3rd-party第三方source ↗×2
20MiniMax M3MiniMax88.990—3rd-party第三方source ↗×2
21GPT-5.6 TerraOpenAI86.387—3rd-party第三方source ↗×2
22GPT-5.6 SolOpenAI85.186—3rd-party第三方source ↗×3⚠±8.4
23Command ACohere8586—3rd-party第三方source ↗×3⚠±48
24MiniMax M2.7MiniMax84.886—3rd-party第三方source ↗×2
25Claude Opus 5Anthropic82.7884—3rd-party第三方source ↗×2
26Claude Sonnet 5Anthropic80.481—3rd-party第三方source ↗
27gpt-oss-120bOpenAI65.866—3rd-party第三方source ↗
28GPT-5.5OpenAI46.39472026-03-11official官方榜source ↗×7⚠±51.61
29Mistral Small 4Mistral AI41.242—3rd-party第三方source ↗
30Claude Haiku 4.5Anthropic39.540—3rd-party第三方source ↗×2⚠±43.5
31Llama 4 MaverickMeta17.818—3rd-party第三方source ↗
32Llama 4 ScoutMeta15.516—3rd-party第三方source ↗

How scores become the Intelligence Score →成绩如何合成智能评分 →