
MiniMax Launches M2.5 Model
MiniMax debuted MiniMax-M2.5 model available on web and APIs. It supports coding, agentic automation, and office tasks in M2.5 and M2.5-Lightning versions via API and MiniMax Agent.
Tag: #launch147 results

MiniMax debuted MiniMax-M2.5 model available on web and APIs. It supports coding, agentic automation, and office tasks in M2.5 and M2.5-Lightning versions via API and MiniMax Agent.

JD.com open-sourced JoyAI-LLM-Flash, a 48B total parameter MoE model with 3B active params, pre-trained on 20T tokens. It excels in knowledge, reasoning, coding, and agents using FiberPO framework and Muon optimizer. Features 1.3x-1.7x throughput gains via dense MTP.

ByteDance's Seedance 2.0 generates lifelike videos rapidly, impressing users worldwide. However, it faces accusations from Disney of copyright infringement for training on protected IP. Generated videos featuring Hollywood celebrities, Disney characters, and comic heroes have proliferated online.

A DIY project named Takumagoko, hand-coded with Claude in just days, has gone viral. It features near-zero costs and ushers in an era of one-sentence hardware creation via AI.

A hand-crafted AI gadget named 拓麻歌子, built using Claude Code in just a few days, has exploded in popularity. The project costs nearly zero and signals the arrival of an era where hardware can be created with a single sentence via AI.

New agentic AI apps interview users for personality matches without swiping. They provide limited, high-compatibility matches based on similarity and reciprocity. Tech entrepreneurs test this love-finding approach.

Jijia Vision launches new VLA large model, shattering embodied AI limits. Achieves near 100% success on long-horizon tasks like folding clothes and making coffee. Handles complex manipulations expertly.

Jijia Shijie's new VLA model breaks embodied AI limits with nearly 100% success on complex long-horizon tasks. It excels at folding clothes, brewing coffee, and folding boxes.