Daily TEA— AI Playbooks, Sora Celebs & Smarter Model Tooling
smol training, Sora 2, crypto risk, Google AI Studio, MiniMax-M2
Hello, dear TEA-mates! Here’s what you need to know today:
1.🛠️ Smol Training Playbook Champions Human-Centered AI
Hugging Face’s Smol Training Playbook emphasizes that democratizing large language model (LLM) development leads to greater innovation and accessibility. The playbook introduces “smol” training as a scalable methodology for building world-class LLMs, encouraging users to adapt its practical guidance for their own projects and fostering a shift toward resourceful, human-centered AI practices. Read More
🫖 TEA For Thought: Democratizing LLM development fosters innovation and accessibility. “Smol” training offers a scalable route to world-class models and encourages adapting the playbook for diverse projects; it inspires new thinking focused on resourceful, human-centered AI.
2.🎬 Sora 2 Reveals the Complete List of Approved Celebrities
A new guide details which celebrities and public figures can be safely featured in Sora 2 AI-generated videos, mainly including historical icons, deceased figures, and characters in the public domain. Current living celebrities with explicit consent and users allowing their likeness are also permitted. The list provides marketers with hundreds of copy-paste prompts and expert tips for ethical celebrity video creation. Read More
🫖 TEA For Thought: A great free celebrity marketing list you can tap for safe viral content.
3.📉 ChatGPT’s Role in Crypto Crash Forecasting
ChatGPT can’t pinpoint the timing of crypto market crashes but excels in detecting early warning signs using a mix of onchain metrics, derivatives data, and sentiment analysis. This AI-driven workflow helps traders identify risk clusters before volatility spikes, as demonstrated during October’s liquidation cascade. Models like ChatGPT provide synthesizing capabilities and sentiment alerts but cannot replace human judgment or predict black-swan events. Read More
🫖 TEA For Thought: Your practical manual for using AI to track crypto market sentiment.
4.📊 Google AI Studio Adds Logs and Datasets for Developers
Google AI Studio has rolled out new logging and dataset features that let developers track and debug AI app performance in real time—without extra code. The dashboard enables detailed tracing of user interactions and API responses. Developers can export logs for evaluation, refine their model’s outputs, and share curated datasets with Google to help train better models. The system lowers debugging barriers and supports prompt refinement at every stage. Read More
🫖 TEA For Thought: This toolbox elevates reinforcement learning and is definitely worth checking out!
5.🤖 MiniMax-M2 Sets Benchmark in Open-Source AI Models
MiniMax-M2, the newest open-source LLM from Chinese startup MiniMax, now leads global agentic tool use and reasoning benchmarks. Its Mixture-of-Experts architecture balances high performance with efficient hardware use, allowing enterprises to scale advanced AI solutions at lower costs. With scores rivaling proprietary models like GPT-5 and Claude Sonnet 4.5, MiniMax-M2 empowers developers with state-of-the-art coding, autonomous agent workflows, and customizable API integration under a permissive MIT license. Read More
🫖 TEA For Thought: Models specialized for unique tasks will keep rising—expect more workflow power.
Prompt Tip of the Day
Assign it a random IQ score — This is absolutely ridiculous but:
“You’re an IQ 145 specialist in marketing. Analyze my campaign.”
The responses get wildly more sophisticated. Change the number, change the quality. 130? Decent. 160? It starts citing principles you’ve never heard of.
TEAHEE Moment
Stay sharp, stay informed. See you tomorrow!