Release feed

New AI tools this week

Oct 24 to Oct 30, 2026. A line appears here only when we have a public date. Missing tools are not old. We just have not confirmed a date.

This week

Oct 30

Try itNew this week

TBC Mischief 1

TBC launched Mischief 1, an image-to-video model on fal and OpenRouter, built on Wan 2.2.

Image and video

TBC press release·fal·OpenRouter

Oct 29

WatchNew this week

Amp

Amp announced macOS Satellites, so an agent can boot a cloud Mac to build and test iOS and Mac apps. It is a waitlist beta.

Vibe coding · AI agents

Amp news·macOS docs·Amp

Oct 29

Try itNew this week

Canva Brush

Canva released Brush, a vision-language model that edits an existing design from a prompt.

Image and video · Productivity

Canva newsroom

Oct 29

Try itNew this week

Mistral Large 4

Mistral published open weights for Mistral Large 4, nicknamed Le Chonk, a 1 trillion parameter multimodal model.

Vibe coding · AI agents

Mistral announcement·Model card

Oct 28

WatchNew this week

Gemini 4 Carbon

Google announced Gemini 4 Carbon, a frontier model for long-horizon coding and cybersecurity. It is not open to developers yet.

Vibe coding · AI agents

Google blog

Oct 28

Try itNew this week

GitHub Copilot

GitHub made Copilot Fusion generally available, so the model picker can route a task to one or more models.

Vibe coding

GitHub changelog·Local models·JetBrains

Oct 28

Try itNew this week

NVIDIA Nemotron 3.5 Transcribe

NVIDIA released Nemotron 3.5 Transcribe, an open-weight speech model for 40 languages, speakers, translation, and audio questions.

Voice and audio · Productivity

Hugging Face blog·Model card·OpenMDW-1.1 license

Oct 28

Try itNew this week

Particle-1

Particle released Particle-1, a coding model post-trained on DeepSeek V4.1 Flash to use fewer tokens per task.

Vibe coding · AI agents

Particle blog·Model docs·Vercel changelog

Oct 27

WatchNew this week

anymove

anymove is an Apache 2.0 pipeline that turns one AI video into a GLB animation for a rigged creature.

Game dev

GitHub

Oct 27

Try itNew this week

Clover-35B-A3B-Preview

DiffuSpace published Clover-35B-A3B-Preview, an Apache 2.0 diffusion language model with 35B total parameters and about 3B active.

Vibe coding · AI agents

DiffuSpace blog·Hugging Face·GitHub

Oct 27

WatchNew this week

Genie 4

DeepMind announced Genie 4, a real-time world model with native audio and dialogue. It is a research preview for trusted testers.

Game dev · Image and video

DeepMind blog

Oct 27

Try itNew this week

LFM3

Liquid AI released LFM3, an open-weight 2B-class model family for text, vision, audio, and on-device use.

AI agents · Voice and audio

Liquid AI blog·Hugging Face

Oct 27

Try itNew this week

Survival LLM

Survival LLM is a free, open source offline assistant with an emergency library that runs on your computer.

Productivity

Survival LLM·GitHub·Hacker News

Oct 27

Try itNew this week

T3 Code

T3 Code v0.1.3 is an MIT-licensed app that runs several coding agents in one place.

Vibe coding · AI agents

GitHub·v0.1.3 release·Theo's launch post

Oct 26

Try itNew this week

Entire

Entire launched a Git forge for agents and humans. Trails replace pull requests, and it is available to every developer.

Vibe coding · AI agents

Entire launch·Pricing·CLI·Seed round

Oct 26

WatchNew this week

Fern 0.1 (Mirendil)

Mirendil announced Fern 0.1, its first model, with a context window of up to 10M tokens. Access is early-access only.

AI agents Vibe coding

Mirendil announcement

Oct 26

Try itNew this week

GPT-6.1 Terra

OpenAI launched GPT-6.1 Terra for computer use, visual reasoning, and writing, in ChatGPT Work, Codex, and the API.

Writing AI agents

OpenAI announcement·API model page

Oct 26

Try itNew this week

Hob

Hob launched as a free, MIT-licensed desktop app that works on your files with local open-weight models, offline.

Productivity AI agents

GitHub·Show HN·Homepage

Oct 25

Try itNew this week

Pomerado

Pomerado launched a service that turns sites without an API into hosted APIs and MCP servers for agents.

AI agents

Launch YC·GitHub·Docs

Oct 24

Try itNew this week

Nanbeige4.3-33B

Nanbeige published Nanbeige4.3-33B, a dense 33B open-weight model for agentic coding, under Apache 2.0.

Vibe coding

Hugging Face·Community GGUF

Earlier releases

October 2026

Oct 23

WatchRelease

Dessa

Block announced Dessa, a full-duplex speech-to-speech model with no public weights or release date.

Voice and audio

Block·TechCrunch

Oct 23

WatchRelease

Diorama

Diorama was announced as an MIT-licensed engine for 2.5D game assets written as TypeScript and exported as sprite sheets.

Game dev Vibe coding

GitHub·Announcement

Oct 23

WatchRelease

Hi3D

Hi3D v1.0, formerly Sparc3D, was released for high-resolution 3D geometry refinement from a coarse mesh and a reference image.

Game dev Image and video

Hugging Face·GitHub·Project page

Oct 23

WatchRelease

Hlas-1

Hlas-1 was posted as a 3.5 billion parameter text-to-speech model with zero-shot voice cloning, under a non-commercial license.

Voice and audio

Hugging Face·Show HN·Demo

Oct 23

WatchRelease

Hy-Nomas

Tencent released Hy-Nomas, an open interactive video world model, with Apache 2.0 weights.

Game dev Image and video

Hugging Face·GitHub·Project page·Tencent Hunyuan post

Oct 23

Try itRelease

Kimi K3.1 Pro

Moonshot released Kimi K3.1 Pro, its first K3.1 model, with a 1 million token context. Full weights are not downloadable yet.

AI agents Vibe coding

Kimi blog·Moonshot forum·OpenRouter listing

Oct 22

WatchRelease

1Password for Agents

1Password launched 1Password for Agents, so approved agents can sign in without seeing your password.

AI agents

1Password blog

Oct 22

Try itRelease

Canva Craft

Canva released Craft, its first in-house image model, in Dream Lab and the public API.

Image and video

Canva newsroom·Canva developers

Oct 22

Try itRelease

Ideogram

New: Ideogram Vector 1.0 (Oct 22, 2026). Ideogram's first vector model makes editable SVGs from a text prompt, or converts an image into an SVG.

Image and video

Vector model page·Launch post

Oct 22

Try itRelease

Ideogram Vector

Ideogram launched Vector 1.0, its first vector model, for editable SVGs from a prompt or an existing image.

Image and video

Vector model page·Launch post

Oct 21

WatchRelease

Claude Duet

Anthropic opened a research preview of Claude Duet, a proactive agent for a morning brief and background tasks.

AI agents

Anthropic article·Product page

Oct 21

WatchRelease

Eleven Realtime Translation

ElevenLabs released Eleven Realtime Translation, a live voice translation model, in alpha for ElevenAPI enterprise customers.

Voice and audio

ElevenLabs announcement

Oct 21

Try itRelease

Flint

Recursive released Flint, an open-weight coding model (about 40 billion total parameters, about 2.5 billion active) under Apache 2.0.

Vibe coding

Recursive announcement·Hugging Face BF16·Hugging Face INT4

Oct 21

Try itRelease

HyperFrames

HyperFrames v0.8.185: an open-source framework that renders HTML to deterministic video.

Image and video

GitHub·HeyGen cloud rendering

Oct 21

Try itRelease

Omnara

Omnara launched an Apache 2.0 managed agent platform you can self-host or run on Omnara Cloud.

AI agents

GitHub·Show HN

Oct 21

WatchRelease

Opposable

Opposable published an Android app that lets a model read the screen and tap and type in other apps.

AI agents

Opposable write-up

Oct 21

WatchRelease

WorldTUI

A research demo showed a 360M-parameter model predicting a terminal screen from keypresses in the browser.

Game dev · world models

Research write-up

Oct 20

Try itRelease

Cursor Agent SDK

Cursor published the Agent SDK, a framework for defining agents in code and deploying them.

AI agents · Vibe coding

Cursor blog·Changelog·Cursor on X

Oct 20

Release

Perplexity

Perplexity Computer can run fully on a supported Apple silicon Mac. It is for Pro, Max, and Enterprise plans. Perplexity recommends 36 GB of memory or more.

Writing Productivity

Oct 20

Try itRelease

Voxtral 3

Mistral released Voxtral 3: speech-to-text, text-to-speech, and a real-time voice agent SDK.

Voice and audio

Mistral announcement

Oct 19

Try itRelease

Gemini API CLI

Google released an experimental preview of the Gemini API CLI for shells and coding agents.

AI agents Vibe coding

Google AI Studio·GitHub

Oct 19

Release

Mercury 3

Inception Labs released Mercury 3, a diffusion language model on its API, OpenRouter, Baseten, and Vercel. The company lists a 1 million token context window.

AI agents

Oct 19

Try itRelease

MiniMax M3.1 Flash

MiniMax rolled out M3.1 Flash, a multimodal model with about 428 billion total parameters and a 1 million token context.

AI agents Vibe coding

MiniMax blog·Hugging Face·API pricing

Oct 19

Release

Viggle JST-2

Viggle introduced JST-2, a physics-native 3D world model for scenes, characters, and motion.

Game dev Image and video

Oct 15

Try itRelease

Kling AI

Kuaishou officially launched Kling 4.0 on October 15, 2026. Native generation runs up to 30 seconds in a single pass.

Image and video

Kuaishou release·Kling 4.0 post

Oct 15

Release

Muse Spark Open

Meta released Muse Spark Open weights. It is a multimodal model aimed at coding and agent work, with a 1 million token context window. Meta says it can run locally on two Apple M5 Ultra Mac Studios.

AI agents

Oct 15

Try itNew this week