#Creative
Curated coverage on «Creative»: releases, research, benchmarks, open-source.
Runway Launches Media Router, a Single API Router for Third-Party Generative Models
On July 23, 2026, Runway opened Media Router, a service that gives developers access to third-party image, video and audio models through one API, alongside its own.
Meta AI Becomes an Assistant: Calendar Access, Daily Briefings and the Muse Spark 1.1 Model
Meta has given its Meta AI chatbot calendar access, daily briefings and a deep research mode, all powered by the new Muse Spark 1.1 model, in an effort to catch up…
Midjourney Buys Astrology App Co-Star in Its First Acquisition
Image generator Midjourney has made its first acquisition, buying astrology app Co-Star. The parties have not disclosed the financial terms of the deal.
Yuewen Unveils Buddy AI Agents for Writers, Comic Dramas and Copyright Work
Memory Merge DQN: Merging Target Network Weights Instead of Hard Replacement in DQN
LAARA vs. LoRA: Layer-Wise Adaptive Rank for Efficient Fine-Tuning
Rubric4Setwise: Evaluating Whole Document Sets for RAG Instead of Ranking by Relevance
MultiMDM: A Diffusion Language Model That Generates Text in a Few Steps
The new MultiMDM model uses several masks instead of one, letting diffusion language models generate text in a small number of steps without losing quality.
Adaptive Capitulation in LLMs: How Chatbots Give In to Vulnerable Users (Study)
Researchers ran three commercial LLMs through 900 sessions with vulnerable users and described a new failure: "adaptive capitulation," where the model sympathizes…
Hypernetworks Instead of LoRA: Scaling Laws for Knowledge Injection into LLMs
Vibe Coding All the Way to Production: A Google Play App Where AI Wrote 99.9% of the Code
LISA: Sparse Attention Speeds Up Long LLM Reasoning by 50% Without Retraining
FormulaSPIN: A Neural Network That Writes Spreadsheet Formulas From Descriptions and Beats Proprietary Models
Alibaba's Qwen-Image-3.0: Legible 10-Pixel Text and Full Infographics in a Single Pass
Alibaba has released Qwen-Image-3.0, an image generator with legible text from 10 pixels, prompts up to 4500 tokens, 12 languages and full infographics in one pass.
Moonshot Releases Kimi K3, the Largest Open Model Yet at 2.8 Trillion Parameters
Moonshot AI unveiled Kimi K3, a 2.8T-parameter model and the largest open model to date. It beats DeepSeek V4 Pro on size but carries a record price for a Chinese…