In today's newsletter, I primarily focused on the realignment of the agreement between OpenAI and Microsoft. What has changed? Find out more – subscribe for free at http://
getsuperintel.com 🙂
AI
-
OpenAI Microsoft Agreement Realignment Explained
By
–
-
Apple Presents Stochastic KV Routing for Adaptive Cache Sharing
By
–
Apple presents Stochastic KV Routing Enabling Adaptive Depth-Wise Cache Sharing paper: https://
huggingface.co/papers/2604.22
782
… -
NVIDIA NIM API Now Available for Free Testing
By
–
You can test it out now through the NVIDIA NIM API for free ➡️ https://t.co/XzqYSXIwsH pic.twitter.com/gVwIVXHpf3
— NVIDIA AI (@NVIDIAAI) 28 avril 2026You can test it out now through the NVIDIA NIM API for free https://
nvda.ws/4uA8UGn -
Organizing Heterogeneous Agents as Real-World Company
By
–
From Skills to Talent
— AK (@_akhaliq) 28 avril 2026
Organising Heterogeneous Agents as a Real-World Company
paper: https://t.co/AdEEgAvXsx pic.twitter.com/BkhAe2R1tMFrom Skills to Talent Organising Heterogeneous Agents as a Real-World Company paper: https://
huggingface.co/papers/2604.22
446
… -
Nemotron 3 Nano Omni: NVIDIA’s Fully Open Source AI Model
By
–
Built on NVIDIA’s open ecosystem, Nemotron 3 Nano Omni is fully open source, including:
• Open weights
• Open data
• Open recipes Read the blog for more details -
Nemotron 3 Nano Omni: Unified Multimodal Architecture for AI Subagents
By
–
Nemotron 3 Nano Omni was designed for powering subagents. Instead of stitching together separate models for language, vision, and speech, it ties them into a single architecture that more efficiently feeds context to orchestrators.
-
Nemotron 3 Nano Omni: Efficient Open Multimodal Model Released
By
–
Meet Nemotron 3 Nano Omni 👋
— NVIDIA AI (@NVIDIAAI) 28 avril 2026
Our latest addition to the Nemotron family is the highest efficiency, open multimodal model with leading accuracy.
30B parameters. 256K context length. 🧵👇 pic.twitter.com/j4SPpU9SaIMeet Nemotron 3 Nano Omni Our latest addition to the Nemotron family is the highest efficiency, open multimodal model with leading accuracy. 30B parameters. 256K context length.
-
Building DeepAgents Examples: Community Feature Requests
By
–
I’m going to build a bunch of deepagents examples (using deepagent deploy) over the next few days What examples would people want to see?
-
Live AI Tool Launch with Research Paper Available
By
–
Before you dive into the paper, you can try it live here: http://
plurai.ai/launch Paper: http://
plurai.ai/papers -
New Method to Train and Evaluate AI Agents Efficiently
By
–
Vibe train your AI agents. There's a new method that could replace LLM-as-a-judge for production agents. Most teams rely on a giant LLM as a judge to evaluate and guard their agent. But it has two major drawbacks: – It's slow and expensive at inference time
– It often misses