LatitudeGames has unveiled Equinox-31B, a storytelling-focused model based on Gemma 31B.
ByteDance has open-sourced a 3B-scale multimodal model with image, video, editing, and reasoning capabilities.
Anthropic is in talks to adopt Microsoft's proprietary AI chip, **Maia**, to secure computing resources.
The article introduces 10 open source tools that are changing the game development pipeline.
xAI recorded a $6.4 billion loss in 2025 while disclosing annual AI infrastructure investment of $30.8 billion.
Herdr was introduced as a terminal workspace that manages AI coding agents in one place.
LangChain unveiled a new architecture that goes beyond token streaming to stream messages, tool calls, and sub-agents as structured events
Microsoft has unveiled MagenticLite and Fara1.5, agentic systems designed for small models.
Concerned about data scraping for AI training, news outlets are blocking Internet Archive's access.
LangSmith Auth Proxy injects authentication headers at the network layer, keeping credentials managed outside the sandbox.
Anthropic has launched 13+ free AI courses covering MCP, Claude Code, and more, complete with certificates.
Runtime has unveiled sandboxed coding agents and control features for teams.
Pinterest introduced a 'One Definition, Many Runtimes' approach to improve the cost efficiency and consistency of its user sequence data platform.
At COMPUTEX 2026, NVIDIA unveiled new AI technologies including the Vera Rubin NVL72 and Jetson Thor.
Dropbox built an internal coding agent platform called **Nova**, tailored to its large monorepo and CI.
Marking five years of accessibility work, GitHub unveiled its strategy and expanded improvements across the entire platform.
Anthropic's Claude Sonnet 4.5 model will become unavailable in chat starting May 26, being replaced by Sonnet 4.6.
Replit has launched a self-service **Replit Enterprise** feature that can be adopted instantly without a separate sales consultation.
OWC has announced a Thunderbolt 5 device that enables local execution of large AI models on Windows and Linux PCs.
This presents the concept of an 'Agent Execution Tax,' where the actual cost per successful task matters more than token pricing, along with benchmark results by model.
A study found that changes in prompt tone can sharply reduce the honesty of small language models.
The 'gtm-mavericks' skill has been released, enabling AI coding agents to build GTM strategy through six marketing expert personas.
ElevenReader has added over 200,000 premium audiobook and ebook titles.
Vega has implemented ZKPs that prove only facts from government-issued credentials without revealing the original documents.