MarkTechPost (1 hour ago)
Edge0 released Audio8-TTS-Realistic, a 0.6B open-weight zero-shot voice cloning TTS model under Apache 2.0. It supports 23 languages and adds a final GRPO stage driven by a learned realism critic.
Read the full story on MarkTechPost.
With full-year capital expenditures projected to hit $220 billion, Amazon’s after-hours trading saw shares retreat roughly 7% despite a blowout quarter.
David Durman, a University of Kentucky administrator, recently earned a religious exemption from using AI in the workplace.
The US Air Force wants artificial intelligence to identify aerial refuelling mistakes, improve pilot training and reduce dependence on human assessments.
Anthropic’s latest copyright lawsuit over Elvis Presley songs exposes the AI industry’s troubling dependence on other people’s creative work.
With Google now injecting “sponsored” links directly into AI follow-up queries, traffic to online publishers will suffer even more.
Nvidia plans $3 billion investment in Meta’s share offering as companies expand their partnership on Muse AI models and infrastructure.
Baseten has released joey-1, a 2B open-weight decision model that returns calibrated probabilities instead of text. It runs at 20 ms P50 latency and costs $0.01 per million input tokens.
Prime Intellect released Prime Mega MoE, a backward pass for DeepSeek’s fused Mega MoE kernel. It makes the kernel trainable and speeds Qwen3 MoE training steps by up to 34% in prime-rl.
Infini-AI-Lab released Hermit, an Apache-2.0 harness adapter framework that trains an external layer of instructions, skills, memory and tools from task feedback, leaving the agent’s model and harness unchanged.
Block’s AI lab released Dessa, a full-duplex speech-to-speech model trained only on human conversation audio. It runs locally on a MacBook Pro M4 Max, but Block has published no size, benchmarks or weights.
71 active sources
26 Turkish · 45 English
All sources
