Meta for Developers

Shovel Up! just went open source, inviting creators to dig in, remix, and build their own worlds. Make your worlds remix-ready here: https://bit.ly/3LrIa9B By becoming a remixable world, Shovel Up! is paving the way for more collaborative, high-performing worlds in Meta Horizon. Our latest blog post covers best practices on mesh combining, asset pooling, and performance testing so you can create your own remixable worlds shine and make a name for yourself as a pioneer in the creator community. Click the link above to get started.

Enterprises are accelerating AI adoption, but many still face fragmented tooling, complex integrations, and strict data and security requirements. The Llama stack provides an open, modular foundation designed to support enterprise-grade AI development and deployment across hybrid environments. Paired with Red Hat AI, organizations gain a consistent platform for model orchestration, security, governance, and scale. Join us to explore: ✅ How open tooling and standardized interfaces simplify production AI ✅ Deployment patterns across hybrid and cloud environments ✅ How Red Hat and Meta are enabling enterprise AI with the Llama stack ✅ Real-world adoption and lessons learned 📅 November 11, 2025 at 1:00 PM ET / 10:00 AM PT Register to attend: https://bit.ly/3Llv51D

Is your world ready for mobile players? Find out in our new tutorial on mobile design best practices from Worlds Creator Academy: https://bit.ly/4hNr4zd Our new “Designing for Mobile” tutorial walks you through best practices for mobile-friendly UI, avoiding common pain points, and creating worlds that feel natural and worth returning to. Click the link above for quick and easy tips to help you spark lasting engagement.

Waiting for the perfect moment to build? The moment just arrived (with $1.5M in prizes). Enter Meta Horizon Start Developer Competition 2025: https://bit.ly/3JDPIpc Whether you’re a solo dev, a small studio, or part of a large team with one big idea, this is your chance to build your part of the next generation of VR. Build with Unity, Unreal, Meta Spatial SDK, Android, React Native, or Immersive Web SDK and compete across gaming, lifestyle, and entertainment tracks. $1.5 million in prizes are up for grabs, so start soon and enter before submissions close on December 9. Click the link above to get started. Eligibility requirements and conditions apply.

When player preferences shifted, Tokyo-based studio MyDearest revamped itself from scratch. 🏗️ Read the studio's comeback story: https://bit.ly/4oWljS8 At the end of 2024, the team faced a harsh truth that players simply weren't seeking out their cinematic VR titles like they used to. When they could have hit the panic button, they pivoted fast. The result was the creation of Bazooka Studio, a development arm built on short cycles and “one concept impact.” Multi-year projects became two-month sprints, and they released 9 games in 8 months. Today, their hit Devil’s Roulette has garnered over 1M organic views on TikTok and a 4.8-star average in the Meta Horizon Store thanks in part to strong player feedback loops and real-time updates. Click the link above to read their story and find helpful tips to help you know (and grow) your audience.

Trying to build a scalable on-premises AI chatbot architecture? Info Edge did it with Llama 3.1 8B. Check out how their student education platform used the compact Llama model and retrieval-augmented generation to run its chat solution more efficiently on GPU infrastructure, while protecting data control and security. Read the full story: https://bit.ly/3WzEhln

# The AI Agents Rule of Two Sharing our latest blog post outlining Meta’s approach to securing AI agents. At Fal.con 2025 in September, I joined a panel on the future of AI security, where I highlighted the important work our teams are doing to address one of the field’s biggest challenges: protecting agentic AI systems from prompt injection and other emerging threats. We’re introducing the “Agents Rule of Two” - a practical framework, inspired by Chromium’s Rule of Two approach, designed to reduce the risk of high-impact security incidents in AI agents. The principle is straightforward: The Rule of Two approach for AI agents helps to ensure agents never have unchecked power - no agent can have a full control loop that could be exploited leading to severe security incidents. This work is the result of deep cross-functional collaboration across Meta and I’m grateful for the expertise and dedication teams brought to this challenge. The “Rule of Two” is already shaping how we build and deploy agentic AI, and will raise the bar for security across the industry. https://lnkd.in/d4HPrXzV

Close 2025 on a high note. Build mobile worlds and enter the Meta Horizon Creator Competition: Mobile Innovation to win a share of $1.5M: https://bit.ly/4oOOC8V Ready to end the year with a win? Our last creator competition of the year challenges you to build mobile-first worlds that push creative boundaries, unveil new interactions, and lean into social features. Enter by December 8 for a chance to win up to 35 prizes, including top prizes of $100K each. Click the link above for competition details.

Every world starts with a creative spark, but it grows through feedback. Start building yours in the Worlds Desktop Editor: https://bit.ly/43RT6DE In our latest Behind the Build spotlight, creator and 3D artist MKE_TheGuru shares how listening to players helps him refine and reimagine popular worlds like Freeze Tag: Sweets & Treats, and provide better support for HabitXR’s builds like Bots! and Plants!. Using the Worlds Desktop Editor, MKE_TheGuru keeps his workflow tight and stays organized when managing multiple projects, organizing assets, and experimenting faster than ever. If you’ve been waiting to start creating, now’s the time. Click the link above to download the Desktop Editor, explore your ideas, and bring your first world to life.

We're excited to announce our upcoming session at the Snowflake Build event on Nov 5 from 1:30 PM - 2:30 PM PST! 🚀 Join Anika Shahi (Partner Solutions Engineer, Snowflake) and Anusha Dudi (Llama Partner Engineer, Meta) in their session: Optimizing Llama on Snowflake: A Cookbook for Performance, Cost, and Architecture http://bit.ly/3X5NEcw