AI HOT Daily · Jun 5, 2026
NVIDIA launched the customizable multimodal safety model Nemotron 3.5 Content Safety, Nex-N2-Pro pushed a 397B MoE reasoning model toward top-tier performance, Google released the real-time music model MRT2, Replit teamed up with Shopify to speed up store creation, ChatGPT introduced its stronger Dreaming memory system, OpenAI APIs added moderation scores, OpenJarvis advanced a local-first personal agent framework, and bot traffic surpassed human web traffic for the first time.
1. Nemotron 3.5 Content Safety
Enterprise-focused multimodal safety model with custom policies, unified text and image review, and auditable reasoning traces.
2. Nex-N2-Pro
A 397B MoE reasoning model based on Qwen3.5, supporting ultra-long context and multimodal input with deeper reasoning at lower token cost.
3. Google MRT2
Google released MRT2, a real-time music model playable through MIDI, text prompts, and gestures with sub-200ms latency.
4. Replit + Shopify
Replit Agent partnered with Shopify to help users build and launch custom stores from natural language prompts.
5. ChatGPT Dreaming
ChatGPT introduced Dreaming, a stronger memory system for better cross-conversation recall and personalization.
6. OpenAI API moderation scores
OpenAI added moderation scores to the Responses and Completions APIs so apps can log, route, review, or block content in the same generation flow.
7. OpenJarvis
Stanford researchers released OpenJarvis, a local-first personal AI agent framework that keeps reasoning, memory, tools, and learning on-device.
8. Cloudflare Radar
Cloudflare Radar reported that bot traffic made up 57.5% of global HTML requests over the past week, surpassing human traffic for the first time.
Sources and verification
This legacy briefing did not retain its original source URLs. Verify safety, financial, policy, and breaking-news claims with authoritative primary sources before acting on them.