Meta AI App 📱, Llama API 💻, Relational Graph Transformers 🌐

Meta has introduced a new standalone AI app, expanding its efforts to integrate AI features into consumer experiences more directly. ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ 

Sign Up |Advertise|View Online TLDR From TLDR AI 2025-04-30 🤖 Together Inference & Fine-Tuning on 200 + OSS Models—One API, Any Scale (Sponsor) Together AI is The AI Acceleration Cloud, delivering turbocharged, production-grade inference – with an integrated code interpreter – as well as fine-tuning for the leading open source models. Migrate from closed models, and utilize serverless or dedicated endpoints without changing your code. 3 key highlights 200 + open-source models – Llama 4 Maverick & Scout, DeepSeek-R1, Qwen QwQ-32B, and more – all available through OpenAI-compatible endpoints, with built-in Code Interpreter. Flexible deployment – Serverless API for pay-per-token usage or Dedicated Endpoints for single-tenant GPUs and consistent performance. Continuous Fine-Tuning of Fully-Owned Models – Evolve your AI models over time, using preference optimization and continued training. Upload your data, run LoRA or full-parameter jobs, and push the tuned model straight to production. Learn more › 🚀Get started with Together Inference and join 550,000+ developers building on open-source models at GPU speed. 💻 🚀 Headlines & Launches Meta AI App (1 minute read) Meta has introduced a new standalone AI app, expanding its efforts to integrate AI features into consumer experiences more directly. Meta API for Llama (4 minute read) Meta revealed a limited preview of its Llama API at LlamaCon, giving developers tools to fine-tune, deploy, and evaluate models from the Llama series. 🧠 Deep Dives & Analysis Language equivariance as a way of figuring out what an AI "means" (5 minute read) A researcher identified syntax-semantics issues in LLMs and proposed language equivariance to address them. The approach suggests checking if LLMs' moral judgments remain consistent after translation. Language-equivariant LLMs could represent a deeper understanding beyond mere syntax. Relational Graph Transformers for Enterprise Data (17 minute read) Relational Graph Transformers can help address enterprise data challenges and power applications like customer analytics, recommendations, fraud detection, and forecasting. AI Companions (9 minute read) AI companions are poised to redefine the digital landscape by evolving beyond simple chatbots into interactive interfaces that provide genuine personalization and engagement. The current generic AI interactions hinder meaningful connections. Future success depends on intuitive designs that make AI feel more like a personal companion rather than a tool. As AI matures, the focus should shift from model quality to enhancing user experience and personalization, turning AI into a true companion that fosters curiosity and self-discovery. 🧑‍💻 Engineering & Research Can you run a full-fledged reasoning model on your local machine? (Sponsor) Imagine building AI apps with advanced reasoning without the cloud bill shock. HP's AI Studio webinar shows how to distill DeepSeek-R1's intelligence into locally deployable models. Learn to enhance Llama 3.1 with chain-of-thought capabilities and run inference entirely on your hardware—no API tokens required. See in action on May 1st CogView 4 Image Generation Model (Hugging Face Hub) CogView 4 is a next generation generative image model that is permissively licensed. It outperforms Flux models on a number of key axes. AI Hedge Fund (GitHub Repo) There have been many attempts to automate trading with modern reasoning models. This one is a bit better than others, although it still needs substantial effort. It uses persona based prompting to aggregate predictions about various fundamental assets. Alias free super resolution (GitHub Repo) This new work can upscale to arbitrary resolutions without typical reconstruction aliasing problems. 🎁 Miscellaneous The 2025 Annual Work Trend Index: The Frontier Firm is born (7 minute read) The Microsoft 2025 Work Trend Index report highlighted the shift toward AI-driven "Frontier Firms," emphasizing intelligence on tap, human-agent teams, and the rise of the "agent boss." Microsoft also announced updates to its Microsoft 365 Copilot that enhance AI integration in workplace applications. OpenAI researcher behind GPT-4.5 denied U.S. green card (2 minute read) A key OpenAI researcher who contributed to GPT-4.5 had their U.S. green card application denied. The decision raises concerns about immigration barriers for top AI talent. Developers increasingly value generative AI expertise (2 minute read) Developers should prioritize generative AI skills for career growth and competitive advantage. Companies are seeking experts to lead AI-driven innovation across projects. TLDR is hiring a curator to join the TLDR AI team (Fully Remote) TLDR is hiring an AI curator to help write our daily newsletter, read by over 700,000 subscribers. We're looking for someone who works full time in AI (at a frontier lab, AI startup, or major tech company) and regularly keeps up with the latest AI news and research. Time commitment is ~5-10 hours/week. To apply please send your LinkedIn or resume to jobs@tldr.tech along with a few sentences on how you currently keep up with AI news! ⚡ Quick Links Accelerate your AI roadmap with a 5-minute self-assessment (Sponsor) Turing's GenAI and LLM Assessment helps AI leads identify blockers in their post-training strategy—from talent gaps to scaling challenges. In five minutes, uncover targeted next steps and resources built for teams leveraging foundation models. Start the self-assessment here NotebookLM Audio Overviews Expanded to 50+ Languages (1 minute read) Google enhanced NotebookLM by enabling its popular Audio Overviews feature in more than 50 languages, allowing wider global access to AI-generated podcast-style summaries. OpenAI wants its 'open' AI model to call models in the cloud for help (3 minute read) OpenAI plans to release a free open-source model that can hand off tasks to its cloud models for better performance. AI radio host debuts in Australia using ElevenLabs tech (2 minute read) The host interacted with listeners and delivered news segments live on air. Love TLDR? Tell your friends and get rewards! Share your referral link below with friends to get free TLDR swag! https://refer.tldr.tech/0b6a6dc1/2 Track your referrals here. Want to advertise in TLDR? 📰 If your company is interested in reaching an audience of AI professionals and decision makers, you may want to advertise with us. Want to work at TLDR? 💼 Apply here or send a friend's resume to jobs@tldr.tech and get $1k if we hire them! If you have any comments or feedback, just respond to this email! Thanks for reading, Andrew Tan, Ali Aminian & Andrew Carr Manage your subscriptions to our other newsletters on tech, startups, and programming. Or if TLDR AI isn't for you, please unsubscribe.