CAMEL-AI.org

The live talk begins in less than 24 hours! Register now and join us live tomorrow! 🔗: https://lnkd.in/eb-qs7Z9

CAMEL Now Supports ERNIE 5.0 & ERNIE 4.5 Turbo VL! Key Features: - ERNIE 5.0: Unified Omni-Modal Modeling: Natively integrates text, images, audio, and video into a single model architecture, enabling seamless cross-modal understanding and generation for complex, multi-sensory agent tasks. - ERNIE 5.0: Efficient MoE Architecture: Features a massive 2.4T+ parameter Mixture-of-Experts design with less than 3% active parameters per inference, delivering frontier-level performance across 40+ benchmarks while significantly reducing computational costs. - ERNIE 4.5 Turbo VL: Advanced Visual Reasoning: Introduces "Thinking with Images" capabilities, allowing the model to zoom into details and perform multi-step reasoning on complex visual data, such as chart analysis and causal relationship interpretation. - ERNIE 4.5 Turbo VL: High-Speed Efficiency: Built on a lightweight 28B parameter MoE architecture (activating only ~3B parameters), offering a perfect balance of speed and intelligence for real-time visual understanding and tool-use scenarios. This integration expands CAMEL-AI's multimodal capabilities, offering developers access to Baidu's latest flagship models for building powerful, efficient, and versatile agentic applications.  Special thanks to Tao Sun for leading the implementation!  Reference: https://lnkd.in/eF74Fbj3

🚨 CAMEL-AI Live Talk: The Context Engineering Techniques at CAMEL Our engineer Hesam Sheikh Hassani will share the thinking behind CAMEL’s memory architecture. What You’ll Learn from this live talk: 1. Context Summarization How to keep agents focused by trimming low-value context. 2. Workflow Memory Give agents reusable “experience” so they get faster over time. 3. Tool Output Caching (Cautionary Tale) Why caching tool outputs looked promising but was rolled back. Join us live and learn how to make your agents smarter and more efficient. 👉 Register here: https://lnkd.in/e-dkFwj4

🐫 Feature Update: CAMEL-AI Now Supports Claude Opus 4.5 Model Key Features: - World-Leading Software Engineering Performance: Claude Opus 4.5 achieves state-of-the-art results on real-world software engineering benchmarks, excelling in complex code generation, multi-system debugging, and autonomous agent workflows with superior reasoning capabilities that handle ambiguity and tradeoffs without requiring detailed guidance. - Enhanced Research and Analysis Capabilities: Significantly improved performance in deep research tasks, document processing (slides, spreadsheets, structured data), and complex problem-solving scenarios, providing more nuanced and context-aware responses for sophisticated multi-agent system development. - Industry-Leading Benchmark Performance: Achieves top-tier results across critical benchmarks including agentic coding (SWE-bench Verified 80.9%), agentic terminal coding (Terminal-bench 2.0 59.3%), agentic tool use (t2-bench Retail 88.9%, Telecom 98.2%), scaled tool use (MCP Atlas 62.3%), computer use (OSWorld 66.3%), novel problem solving (ARC-AGI-2 37.6%), graduate-level reasoning (GPQA Diamond 87.0%), visual reasoning (MMMU 80.7%), and multilingual Q&A (MMMLU 90.8%), outperforming comparable models across diverse evaluation metrics. This integration expands CAMEL-AI's model ecosystem with Anthropic's most capable model, providing developers with cutting-edge AI capabilities for building sophisticated autonomous agents and complex software engineering applications. By leveraging Claude Opus 4.5, we enabled the CAMEL Agents to autonomously build and store an interactive Rubik’s Cube webpage locally. Special thanks to Wendong Fan for implementing this integration! Reference: https://lnkd.in/eMx-Av5y

Great to collaborate with Dify and Amazon Web Services (AWS) on the London Meetup! During the session, our founding members Douglas Lai and Yuqin Xie presented “Building the Future of Multi-agent Workforce,” introducing the CAMEL framework through a focused look at the evolution of agents, from early automation to modern LLM-driven systems capable of advanced task execution, world simulation, and data generation. We shared the core motivations behind CAMEL’s design and discussed how a multi-agent workforce addresses the key challenges in building a reliable, scalable, and well-coordinated agentic future. To make this vision concrete, we showcased three representative use cases demonstrating how Eigent AI, brings multi-agent systems into real-world production environments and how our open-source ecosystem bridges the gap between academic research and industrial adoption. 🙌 Thanks to everyone who came by!

Exciting updates: CAMEL-AI Now Supports Gemini 3 Model! 👉 Most Intelligent Model Yet: Integration of Gemini 3 Pro model, featuring breakthrough performance across reasoning, mathematics, and multimodal tasks with top benchmark scores. The model delivers exceptional capabilities in coding, vision processing, and complex problem-solving. 👉 Massive Context and Advanced Reasoning: Support for Gemini 3's 1 million-token context window enabling unprecedented long-form processing capabilities, combined with advanced reasoning through Gemini 3 Deep Think mode for tackling complex multi-step problems requiring extended contemplation and sophisticated analysis. 👉 Superior Agentic and Coding Capabilities: Native integration with Gemini 3's exceptional agentic abilities, scoring 54.2% on Terminal-Bench 2.0 and 76.2% on SWE-bench Verified, with top performance in WebDev Arena (1487 Elo). The model excels at zero-shot generation, complex prompts, and autonomous task execution with improved long-horizon planning. This integration brings Google's most advanced AI model to CAMEL-AI, offering developers cutting-edge reasoning, coding intelligence, and agentic capabilities for building next-generation autonomous multi-agent systems with unprecedented context understanding and problem-solving abilities. As a demonstration, the CAMEL Agent powered by Gemini 3 generated an interactive Rubik's Cube webpage and saved it locally.

CAMEL-AI 🐫 Feature Update: Browser Sheet Interaction with Input and Read Capabilities Key Features: 👉 Spreadsheet Input Operations: Enhanced HybridBrowserToolkit with browser_sheet_input functionality using relative positioning algorithms, enabling precise data entry into spreadsheet cells with intelligent cell targeting and batch input operations for efficient data manipulation. 👉 Sheet Reading Capabilities: Introduces browser_sheet_read method with original position label support, allowing agents to extract and interpret spreadsheet data while maintaining cell reference integrity and supporting complex data structures across multiple sheet formats. 👉 Batch Keyboard Input System: Implements advanced batch keyboard input functionality across TypeScript and Python layers, enabling efficient multi-cell operations with optimized input sequences and reduced interaction overhead for large-scale spreadsheet automation. 👉 High-Level Action Tracking: Adds comprehensive action tracking decorator with error handling and context management, providing detailed logging and monitoring of spreadsheet interactions while ensuring robust error recovery and operation consistency. This enhancement enables sophisticated spreadsheet automation workflows, supporting complex data entry, extraction, and manipulation tasks through intelligent browser-based interactions with popular spreadsheet applications. Special thanks to puzhen zhang for leading the development and implementing the comprehensive spreadsheet interaction features! Reference: PR #3405 — Feat: browser sheet with input and read capabilities https://lnkd.in/eWA9wD7Y

CAMEL-AI Feature Update: Gmail Toolkit Integration Key Features: 👉 Comprehensive Email Management: Complete Gmail integration with 19 essential tools covering email sending, receiving, drafting, replying, forwarding, and thread management, enabling agents to handle complex email workflows with full OAuth2 authentication and automatic token refresh. 👉 Advanced Message Operations: Sophisticated email processing capabilities including attachment handling, label management, message filtering with Gmail search syntax, and intelligent HTML/plain text body extraction with automatic content parsing and formatting. This integration transforms CAMEL-AI agents into powerful email assistants capable of managing complex communication workflows, from simple message sending to sophisticated email automation with full Gmail feature support. 👏 Special thanks to Waleed Alzarooni for leading the development, and Wendong Fan and Xiaotian for their thorough reviews and enhancements!