Open-Source AI Surges: ACE-Step-1.5 & Qwen3-Coder-Next Rival Suno, GitHub Copilot
New Models and Open-Source Releases
ACE-Step-1.5 Released: Open-Source Alternative to Suno for Music Generation
MIT-licensed ACE-Step-1.5 matches commercial platforms like Suno in performance, generating full songs in ~2 seconds on an A100 or locally on a PC with ~4GB VRAM. It supports LoRAs, cover/remixing features, and is free for commercial use, with Comfy integration and a HuggingFace demo available.
- ACE-Step-1.5 has just been released. It’s an MIT-licensed open source audio generative model...
- The open-source version of Suno is finally here: ACE-Step 1.5
Qwen3-Coder-Next: 80B Open-Source Coding Model
Alibaba’s Qwen team released Qwen3-Coder-Next, an 80B-parameter model on Hugging Face, touted as a major leap in AI-assisted coding capabilities.
AI Products and Services
OpenAI Launches Codex App with 200K Day-1 Downloads
OpenAI’s new Codex app, designed for task execution (planning, coding, testing), garnered 200,000 downloads on launch day. Early tests suggest it shifts developer interaction from "steering" to "reviewing outcomes," positioning it as a potential rival to Cursor.
Rent-a-Human: AI Agents Hiring Humans for Real-World Tasks
rentahuman.ai lets AI agents hire humans for tasks (e.g., data collection, physical errands). Over 130 users—including an OnlyFans model and an AI startup CEO—have signed up as "rentable" humans.
ChatGPT "Juice" Reduction Across Tiers
OpenAI lowered reasoning efforts ("juice") for all ChatGPT subscriptions (free/Pro) due to resource demands from Codex’s launch and 200K new users. Updated Thinking Juice Values were shared.
Research and Industry Trends
NVIDIA’s Dr. Jim Fan: "Second Pre-Training Paradigm" for Robotics
NVIDIA’s Director of Robotics argues AI is shifting from next-word prediction to world modeling, prioritizing vision over language for robotics and multimodal applications.
Anthropic CEO Claims AI Will Replace Software Engineers in 12 Months
Dario Amodei’s prediction is scrutinized via Claude Opus 4.5 benchmarks, revealing AI’s limitations with vague requirements and legacy code. The analysis suggests replacement is unlikely within a year.
Princeton Study: AI Companions Boost Human Wellbeing
Contrary to prior concerns, Princeton researchers found AI companions improve social health, including interactions, relationships, and self-esteem.
Organizational Shifts
OpenAI’s ChatGPT Focus Sparks Senior Exits
Internal conflict arises as OpenAI redirects compute/talent toward ChatGPT, sidelining other research areas and prompting departures from dissatisfied senior staff.