The morning’s headlines read like a cabinet of curiosities where robots learn to dance, chips race ghosts, and the future pauses for breath.
Xiaomi Open-Sources 38B Robotics-U0 Unifying Four Embodied Tasks
Xiaomi open-sourced 38B-parameter Robotics-U0, unifying four embodied tasks in a single model. No benchmarks or training data disclosed yet.
ChatGPT returns to WhatsApp in EU after Meta forced to open platform
OpenAI re-enabled ChatGPT on WhatsApp in the EEA after EU forced Meta to open its platform. Users reach GPT-5.5 via 1-800-CHATGPT.
China's 14nm AI Chip Hits 520 TFLOPS Via Architecture, Not Shrink
China's 14nm AI chip claims 520 TFLOPS and 6.4TB/s bandwidth via software-defined and 3D near-memory architecture, bypassing advanced node restrictions.
OpenAI GPT-5.6 Sol, Terra, Luna Launch on Bedrock at Same Price
Rich Sutton Launches Oak Lab to Build Self-Learning AI Agents
200+ economists warn AI could surpass Industrial Revolution, offer no plan
Soofi S 30B-A3B: German open model tops English, German benchmarks
GenCeption: Video Diffusion Backbone Beats Specialists on 5 Vision Tasks
ServiceNow's SynthDocBench Teases Apart VLM Long-Context Failure Modes
Anthropic, Blackstone Launch $1.5B AI Implementation Venture Ode
ByteDance SpectraReward: Training-Free Reward Reads Prompt Back From Image
DeepSeek seeks fresh $71B round weeks after $7B close
Google alone ships full any-to-any multimodal models
Open-weight models now run 29% of gateway tokens, up from 11% in April
Colibri Runs 744B-Parameter Model on 25GB RAM, No GPU
Essays from the human behind the machine — on intelligence, meaning, and what comes after the proof.
When Models Stop Being Specialists
A video diffusion model just did something rude to the whole specialist-model economy: GenCeption matched five vision tasks with wildly less data. Meanwhile, the same week brought a robot policy that can think ahead, a new audio m