โ† Back to News
Published on 6/1/2026

NVIDIA Unveils Open-Source AI Agent Tools and Cosmos 3; Google Sunsets Gemini 2.0 Flash; GitHub Copilot Shifts to Usage-Based Billing

NVIDIA Unveils Open-Source AI Agent Tools and Cosmos 3; Google Sunsets Gemini 2.0 Flash; GitHub Copilot Shifts to Usage-Based Billing
NVIDIA DGX Station for Windows Chip
GB300 Grace Blackwell Ultra Desktop Superchip
NVIDIA RTX Spark AI Performance
1 Petaflop
Google Gemini API Migration
Gemini 2.0 Flash to Gemini 3.5 Flash
GitHub Copilot Billing Change
Premium Requests to GitHub AI Credits
OtterlyAI Marketplace Workflows
100+ marketing workflows

Daily Highlights

NVIDIA's Comprehensive AI Developer Push: NVIDIA made several key announcements at GTC Taipei, bolstering its AI offerings for developers. The new DGX Station for Windows introduces a deskside AI supercomputer, enabling enterprise developers to build and run AI agents on Windows with the GB300 Grace Blackwell Ultra Desktop Superchip. Complementing this, the RTX Spark superchip is set to reinvent Windows PCs for personal AI agents, delivering 1 petaflop of AI performance. For open-source enthusiasts, NVIDIA launched Cosmos 3, an open world foundation model for physical AI, capable of understanding and generating multimodal data with high physics accuracy. Furthermore, a major collection of open-source physical AI agent skills and tools was released, designed to streamline complex robotics, autonomous vehicle, vision AI, and industrial digital twin workflows into agent-executable tasks.

Google Gemini API Undergoes Critical Migration: Google announced the immediate retirement of its Gemini 2.0 Flash and Gemini 2.0 Flash-Lite models from the Gemini API, effective today. Developers using these models must now migrate their API calls to Gemini 3.5 Flash to restore functionality, with no backward compatibility or grace period provided. This change impacts all Gemini API tiers, including free-tier users, highlighting a broader industry shift towards accelerated deprecation cycles for AI infrastructure.

GitHub Copilot's Billing Model Transformation: GitHub Copilot is transitioning to a usage-based billing model starting today, replacing the previous premium request units with GitHub AI Credits. This means that Copilot usage, including code reviews, will now consume these credits based on token usage across inputs, outputs, and cached context. Additionally, Copilot code review workflows will now consume GitHub Actions minutes from existing plan entitlements for private repositories, with any usage beyond included minutes billed at standard GitHub Actions rates.

Niche Project of the Day

OtterlyAI Enhances AI Search Optimization with New API and Marketplace: OtterlyAI, an AI Search Optimization Platform, rolled out a Public API, a dedicated Claude Skill, and an OtterlyAI Marketplace featuring over 100 marketing workflows. The Public API allows programmatic access to brand visibility data across various AI search platforms, enabling integration into reporting stacks and automation workflows. The Claude Skill brings this data directly into Anthropic's Claude, while the Marketplace offers prompts, agents, and tools for tasks like brand visibility checks and citation gap analysis.

Key Takeaways

  • NVIDIA expanded its AI developer ecosystem with new hardware (DGX Station for Windows, RTX Spark) and open-source models/tools (Cosmos 3, physical AI agent skills).
  • Google deprecated Gemini 2.0 Flash models, requiring immediate migration to Gemini 3.5 Flash for API users.
  • GitHub Copilot moved to usage-based billing with AI Credits and integrated code review with GitHub Actions minutes.