AI in week 33.

G5

GLM 5.3

Model · Z.ai

8

Z.ai's open-weights 753B MoE language model, a post-trained GLM-5.2 aimed at coding, agentic tasks and cybersecurity.

G4

Grok 4.7

Model · xAI

5

xAI's Grok 4.7 language model for coding and knowledge work, available via API, Cursor and Grok Build.

G4

Grok 4.6

Model · xAI

5

xAI's Grok 4.6 closed LLM for reasoning and coding, available via API, Cursor and Grok Build.

Q3

Qwen 3.8 27B

Model · Alibaba

3

Compact 27B-parameter model from Alibaba's Qwen family, runnable locally on a single high-end consumer GPU.

MG

Muse Glimmer

Model · Meta

3

Meta's open-weights 30B dense model for local agents, tuned for tool use and long tasks, Apache 2.0 licensed.

GU

GPT Ultrafast

Model · OpenAI

2

OpenAI API service tier running GPT-5.6 Sol up to 14x faster than Standard, at up to 750 tokens/sec on Cerebras hardware.

G3

Gemini 3.7 Flash

Model · Google

2

Google's fast, lower-cost multimodal Flash-tier Gemini model for the Gemini app, AI Studio and coding tools.

DE

DeepSeek-V4-Pro

Model · DeepSeek

2

DeepSeek's V4 Pro large language model, available via API and as an open model, with low per-token pricing.

N3

Nemotron 3.5 Lightning

Model · NVIDIA

2

NVIDIA's open-weight 30B mixture-of-experts model with 1M-token context, aimed at long-running agent workloads.

CN

Cactus Needle 2

Model · Cactus

1

Tiny on-device tool-calling and extraction model from Cactus, shipped as an 8–29 MB binary for phones, wearables and microcontrollers.

CW

Claude watermarking

Research · Anthropic

1

Anthropic's embedded text watermarks and C2PA Content Credentials for Claude output, built to meet EU AI Act Article 50(2).

CA

Claude AI Content Watermarks

Research · Anthropic

1

Anthropic's machine-readable marking of Claude output: embedded text watermarks and C2PA Content Credentials on files.

MA

MAI-Image-2.6

Model · Microsoft

2

Microsoft AI's image generation and editing model, with a lower-cost Flash variant.

Community ComfyUI workflows that use the MiniMax H3 video model with a one-frame count to output still images.

L2

LTX 2.5

Model · Lightricks

3

Open-weights Lightricks video model with native audio, multi-shot generation, auto duration and up to 4K output.

JV

JoyAI Video Edit

Model · JD

1

Open-weights 16B autoregressive diffusion model from JD for real-time, instruction-guided editing of streaming video.

MA

MAGI-2

Model · Sand AI

1

Open-weights 114B MoE model from Sand AI that generates synchronized video and audio from text or image prompts.

WA

Wan3.0

Model · Alibaba

1

Alibaba's Wan video generation model, reported to produce clips up to about 30 seconds, available via the Wan platform.

Community-quantized W4A8 and int8 VAE checkpoints of MiniMax H3 video model for ComfyUI, by Kijai.

MH

MiniMax H3 Turbo LoRAs

Model · LightX2V

1

Few-step LoRA for MiniMax-H3 that cuts audio-video sampling from about 20 steps to 4–8 in ComfyUI.

MI

MiniMax-H3-Realism-People-LoRA

Model · fal

1

LoRA adapter from fal for MiniMax H3 that targets realistic people, faces and natural skin in text/image/reference-to-video.

IT

Index TTS 2.5

Model · IndexTeam

1

Open-weights zero-shot text-to-speech model from Bilibili's IndexTeam with voice cloning and emotion control in 5 languages.

MI

MiDashengLM-Gen

Model · Xiaomi

1

Xiaomi's LLM-driven model that generates mixed audio scenes (speech, music, SFX, ambience) from structured text captions.

WF

Wispr Flow

App · Wispr

1

Voice dictation app that turns speech into edited text in any app on Mac, Windows, iOS and Android, with an AI notetaker.

MH

MiniMax H3 audio generation

Research · MiniMax

1

Open-weight multimodal model from MiniMax whose video pipeline can be run audio-only to generate music and sound effects.

MM

MiniMax Music 3

Model · MiniMax

2

Open-weights text-to-music model from MiniMax that generates songs with vocals and lyrics across genres and languages.

SU

Suno

App · Suno

2

AI music generator that creates full songs with vocals and instrumentation from text prompts or lyrics.

Open-source local AI model for generating music loops, with reference-loop stacking and MIDI conversion.

A1

ACE-Step 1.5 XL

Model · ACE Studio

1

Open-source local music generation model with a 4B-parameter DiT decoder, extending ACE-Step 1.5 with higher audio quality.

Open-source model that transcribes songs into separate instrument tracks and MIDI notes, including vocals.

HW

Hunyuan3D WorldClaw

Research · Tencent

2

Tencent Hunyuan research framework that turns one text prompt into an explorable, editable open-world 3D scene.

GR

Grokbot

Agent · xAI

9

Grok Bot is xAI's personal and team AI agent, run on cloud VMs for scheduled tasks, X research and phone calls.

CC

Claude Chrome Extension with Cowork

App · Anthropic

1

Claude in Chrome side panel now runs a Claude Cowork session shared with Claude's desktop, web and mobile apps.

CC

Claude Code Session Messaging

App · Anthropic

1

Claude Code feature that lets separate Claude Code sessions send each other messages and summaries to share context.

Microsoft's coding-focused AI model in the MAI family, announced with few public details.

CC

Claude Code Auto Mode

App · Anthropic

1

Claude Code permission mode where a classifier approves safe tool calls and blocks risky ones instead of prompting.

TL

The Librarian 2

App · Matt Wolfe

1

Browser-based 3D roguelite game by Matt Wolfe, built with AI-assisted "vibe coding"; sequel to The Librarian.

CC

ChatGPT Computer History

App · OpenAI

2

Opt-in OpenAI feature that turns activity across apps and websites into memories and a timeline for ChatGPT and Codex.

CD

ChatGPT Desktop App for Linux

App · OpenAI

1

Native ChatGPT desktop client for Linux from OpenAI, bundling ChatGPT, ChatGPT Work and Codex.

D2

Dyna 2

Model · Dyna Robotics

1

World-action model from Dyna Robotics pre-trained on over 1 million hours of egocentric human video for robot manipulation.

SL

Sign language to text model

Model · Google DeepMind

1

Google DeepMind model that translates sign language from a phone camera into text, starting with American Sign Language.

MA

MatrAIx

Research

1

Simulated-user platform with 8.3 billion persona agents for evaluating personal agents, plus RL environments and SFT data.

Google DeepMind research model that transcribes sign language from video into text captions.

TF

The Future Is For Everyone (Meta personal superintelligence essay)

Research · Meta

1

Mark Zuckerberg's essay on Meta's philosophy for building and distributing personal superintelligence to everyone.

NVIDIA open-source model router that selects a suitable LLM for each task to reduce cost.

CO

ComfyUI-MiniMaxH3-Easy

Library · nkxx188

1

ComfyUI custom node suite that simplifies MiniMax H3 video and image generation in a single compact workflow.

MH

MiniMax H3 GGUFs

Model · Unsloth

1

GGUF quantizations of MiniMax-H3 video-with-audio generation model, for stable-diffusion.cpp and Unsloth.

Small 2D tiny VAE for live latent previews of MiniMax-H3 video generation in ComfyUI via KJNodes.