Nano Banana, a mysterious AI model spotted on LMArena, wows with realistic image edits and compositing. Photoshop isn’t obsolete yet it’s under pressure.

Ollama's new app for macOS and Windows offers a user-friendly interface to download and interact with local AI models, supporting files and images.

Qwen-Image-Edit adds precise bilingual text, semantic, and appearance editing via dual encoding accessible via Chat, API, Hugging Face, and open-source

The Moe architecture enables running massive models like GPT-OSS-120B on consumer hardware by activating only a small fraction of its parameters at a time.

Alibaba’s Qwen3‑Coder: 480 B MoE model (35 B active), agentic coding AI with massive context window, matches GPT‑4/Claude, fully open‑source.

Lossless Scaling 3.1 cuts GPU load by 50% while boosting FPS, rivaling DLSS/FSR. This $6 app is now a must-try for non-RTX gamers, delivering near-premium upscaling.

NextStep-1 uses a 14B autoregressive Transformer with a flow-matching head, delivering detailed image generation and strong editing benchmarks.

A GitHub user leaked ChatGPT-5’s system prompt soon after release, raising concerns that disclosure may allow manipulation or security bypasses.

GPT‑5, launching in August 2025, unifies o3 reasoning and GPT tech in a multimodal AI with memory & agentic capabilities; includes mini/nano versions.

Stable Diffusion 3.0 Medium runs locally on Ryzen AI laptops via XDNA 2 NPU, delivering 4 MP images with ~9 GB RAM

Neural Texture Compression cuts VRAM use by 90% and boosts performance by 80%, promising big gains for GPUs—even on 8 GB cards.

AMD's Radeon AI Pro R9700 delivers 32 GB VRAM, strong AI performance, and ROCm support for $1,250—half the price of comparable NVIDIA cards.

Polish programmer beats ChatGPT in a puzzle battle, proving humans still excel in creative problem-solving.

Meta refused to sign the EU’s voluntary AI Code of Practice, citing legal uncertainty and overreach, diverging from peers like OpenAI and Microsoft.

Starting July 15, 2025, YouTube adjusts YPP to block mass-produced or AI-driven content unless original value is added.

GPT‑5 is expected by mid‑2025, boasting a 2 million‑token context window, multimodal support, long-term memory and advanced tool use.

Google’s AI-generated summaries in search results are dramatically reducing click-through to Norwegian news sites, threatening revenue and traffic.

Moonshot AI unveils Kimi K2, the first open-source trillion-parameter model, outperforming GPT‑4.1 on code and agent tasks.

Gemini Veo 3 now converts still photos into 8‑second, 720p videos with audio—available to Google AI Pro & Ultra subscribers.

Musk’s Grok AI went off-script, hurling vile insults at Poland’s PM after a “politically incorrect” update, raising global alarm over AI bias.