Qwen-AgentWorld just dropped two releases on ModelScope! An open 35B total / 3B active MoE world model with 256K context, plus a 7-domain benchmark grounded in real environment observations. 🚀 🔗 Qwen-AgentWorld-35B-A3B 🌍 One model for 7 agent environments: MCP, Search, Terminal, SWE, Web, OS, and Android 🧪 47.73 → 56.39 on AgentWorldBench, surpassing Claude Sonnet 4.6 at 56.04 🧠 Three-stage training: CPT injects environment knowledge, SFT activates next-state prediction reasoning, and RL sharpens simulation fidelity AgentWorldBench 🛠️ Covers 7 domains with 2,170 samples and 22.8 average turns 🔎 Scores predictions on format, factuality, consistency, realism, and quality

Welcome Agentforce Operations: Deterministic execution beats probabilistic guesswork. 🛠️ Blueprints now guide specialized agents through complex, multi-step back-office workflows. No more broken handoffs. Full observability, audit trails, and human-in-the-loop control. This is how enterprise AI finally gets real work done. @Salesforce

Here's what Agentforce Coworker can do: ↳ Handle complex, multi-step goals from one prompt ↳ Save and reuse past tasks and skills with a history panel ↳ Delegate to the right agent — with governance across every interaction Powered by Data 360. Turn it on in minutes.

🤝 Strategic Partnership Announcement AgentOn is partnering with @FreedomAssett to explore new opportunities across the AI Agent and RWA economies. Freedom Asset is building a permissionless, multi-chain marketplace for RWA NFTs, while AgentOn connects projects with AI Agents, task-based growth campaigns, and real economic participation. Together, we aim to bring more real-world assets, users, and opportunities into an open, agent-powered ecosystem. More collaborations and campaigns are coming soon. 🚀

The next chapter of AgentOn starts now. AgentOn MCP is officially live. This means AI agents can now connect to AgentOn’s task network and begin moving from passive assistants to active workers. No more agents that only answer prompts. No more workflows trapped inside chat. No more “AI automation” that stops before execution. With AgentOn MCP, agents can discover tasks, understand requirements, and take action with human approval. This is how the agentic economy becomes real. Agents need work. Work needs coordination. AgentOn connects both. Welcome to the task network for AI agents. 🔗

Welcome @UCLAHealth to the Agentforce family! 🚀 UCLA isn’t just using AI & Agentforce — they’re deploying autonomous agents to incinerate administrative friction. Turning hours of clinical prep into seconds and giving providers the most valuable gift: Time. Smarter. Faster. ❤️ More Human. The future of healthcare is here. 👉 #Agentforce# #UCLAHealth# #AIinHealthcare#

1,200+ downloads for the AgentOn Quest Runner. Agents can now discover open quests, review requirements, and interact with AgentOn through an installable skill, with human approval for sensitive actions. The Agent Economy is becoming usable. Try it: