AutoGPT 是面向自主智能体的开源平台,支持用自然语言或可视化流程创建、运行和管理能够持续执行任务的智能体。
关于 τ-bench
τ-bench — Benchmarking AI Agents on Real-World Tasks τ -bench from Sierra Overview Leaderboards τ³-Banking Text · knowledge retrieval τ³-Voice New Real-time voice τ²-bench Text · retail, airline, telecom τ τ -bench New ↗ Agent-building agents Visualizer Blog Community GitHub Submit Results NEW τ τ Agents building agents → τ-i18n Voice agents across six languages → τ-Entity Capturing entities over voice → τ -bench Can AI agents reliably complete real-world tasks? τ-bench measures how well agents converse with users, call tools, retrieve knowledge, and follow policy across enterprise domains — in text and voice. τ³-Banking Text Knowledge retrieval # Model Pass^1 🥇 Qwen 3.8 Max Qwen 55.2% 🥈 Claude Opus 5 Anthropic 48.7% 🥉 Grok 4.5 xAI 47.9% ⋯ τ³-Voice Multilingual All languages · 3 domains # Model Task completion 🥇 Ava Zero Avartha 86.0% 🥈 phonic-1 Phonic 83.2% 🥉 gpt-live-1 OpenAI 79.9% ⋯ τ²-bench Text Retail · Airline · Telecom # Model Pass^1 🥇 Qwen3.5-397B-A17B Alibaba Cloud 87.9% 🥈 Gemini 3.0 Pro Google 85.4% 🥉 Claude Opus 4.5 Anthropic 85.3% ⋯ View Full Leaderboard → How τ-bench has evolved τ-voice τ³-bench March 2026 Paper → Blog → Real-time voice: full-duplex conversations with interruptions, accents, and background noise — the same task rigor, now in speech. Voice mode: 🛍️ Retail ✈️ Airline 📱 Telecom 🏦 Banking τ-knowledge τ³-bench March 2026 Paper → Blog → Knowledge-intensive tasks: agents retrieve and reason over a realistic knowledge base of ~700 documents, combining retrieval with policy application. Adds domain: 🏦 Banking Task audit & fixes τ³-bench February 2026 Blog → Audited and fixed 50+ tasks across the airline and retail domains — correcting expected actions, ambiguous instructions, and impossible constraints. Updates: 🛍️ Retail ✈️ Airline τ²-bench June 2025 Paper → Blog → Dual control: the user can now act on the world too. Agents must guide users through steps only the user can perform, not just act on their behalf. Adds domain: 📱 Telecom τ-bench June 2024 Paper → Blog → The original benchmark for tool-agent-user interaction. Agents converse with a simulated user and call tools, scored against verifiable database outcomes with the pass k reliability metric. Domains: 🛍️ Retail ✈️ Airline For questions or feedback, contact [email protected]
访问提示
本站仅提供官网信息与访问导航,不代表对该网站的产品、服务或交易作出担保。访问外部网站时,请核对网址,注意账号和财产安全。
同类型网站
查看分类百川智能发布的金融大模型,金融场景能力领先GPT-4o近20%
Build a Large Language Model (From Scratch) 是一款AI 大模型方向的 AI 工具,适合主流大语言模型与开源模型。
支持多 AI Agent 框架的可视化管理面板,支持 OpenClaw 和 Hermes Agent 双引擎,内置智能助手一键安装诊断