AI News
Beni
Jun 11, 2026
Mondo Robotics unveiled Beni, an all-terrain camera robot whose onboard perception and IMU track people or pets at up to 18 mph, stabilize 4K video, and automatically edit highlight reels on the move.
Sports AI
Jun 12, 2026
FIFA deployed optical tracking, a 500Hz connected ball, player 3D scans, and AI-generated avatars for offside detection, while Football AI Pro gives all 48 World Cup teams match analytics for every match.
Autonomous Driving
Jun 16, 2026
Mobileye announced a vertically integrated robotaxi fleet using its self-driving stack and Moovit app, starting with 100 vehicles in a U.S. city in 2027 and targeting about 17,000 over five years.
Robotics AI
Jul 9, 2026
Mistral released Robostral Navigate, an 8B embodied navigation model that follows language commands using one RGB camera, scoring 76.6% on unseen scenes after simulation training across 6,000 environments.
OpenAI Models
Jul 9, 2026
OpenAI released GPT-5.6 Sol, Terra, and Luna, adding stronger coding, computer use, design, science, and cyber performance plus max and multi-agent ultra modes for harder long-running tasks.
AI Resources
Open Source LLM
Tencent open-sourced Hy3 under Apache 2.0, using sparse activation to run with less than half the hardware of larger rivals while showing strong web research and tool-use results in its tests.
World Model
Kyutai and General Intuition open-sourced MIRA, a 5B world model that generates synchronized four-player Rocket League at 20 fps from video and actions, with code, training data, and a demo.
Model Training
Thinking Machines Lab and Bridgewater fine-tuned Qwen3-235B on expert-labeled finance examples, reaching 84.7% accuracy versus roughly 50% for frontier models while cutting task cost 13.8x in testing.
AI Events
AI Summit Barcelona
Barcelona ⏰ Sept 22-23
An experiential event featuring keynotes and working sessions that highlight European AI innovation and industry-leading builders.
Ai4 2026
Las Vegas ⏰ August 4-6
Recognized as one of the largest AI conferences in the U.S., this event connects technology leaders and business executives to showcase proven AI best practices across industry sectors.
What’s New at BasicAI
Blog Pick
How to Write Image Annotation Guidelines for Computer Vision Projects
Read: Learn how to write image annotation guidelines for computer vision projects, covering task scope, ontology, tools, QA, and version control.
CVPR 2026 Top Papers Breakdown: Award Winners and Notable Works
Read: Explore the CVPR 2026 Best Paper, Best Student Paper, honorable mentions, and notable works. See where computer vision AI is heading.
Social Media Highlights
LinkedIn: GPIC is a 100M-image, open dataset from Stanford designed to advance open visual generation research.
Facebook: VideoCUPS is an unsupervised video panoptic segmentation method using ordinary monocular videos.
Subscribe to receive monthly AI newsletter
