Candidates are discovered daily from Hacker News (Show HN), new GitHub projects and trending Hugging Face Spaces, and only appear here after human review. 0 candidates are currently in the review queue.
Google's video model Gemini Omni 1.2 Flash is now generally available: clips up to 15 seconds, multi-turn conversational editing of the same clip, and better adherence to reference images and audio. Gemini API pricing is $0.07 per second at 720p (30% lower than 1.1). 15-second videos in the Gemini app are rolling out first to Google AI Ultra subscribers.
Source: Google Blog · New tools have not been tested by long-term use. Evaluate them yourself.
OpenAI introduced ChatGPT Images 2.6, a new version of the image maker inside ChatGPT that can also create short animated GIFs, plus two new image API models. It is rolling out gradually, so you may not see it right away. Plan availability and pricing: to be verified.
Source: OpenAI · New tools have not been tested by long-term use. Evaluate them yourself.
Kino (YC S23) relaunched as a desktop video editor and media manager where your team and agents such as Claude Code and Codex work on the same timeline. Footage is transcribed and searchable, and pro formats like ProRes, BRAW and R3D are supported. The full editor is free; Pro is $25 per editor per month. Desktop only for now, no browser version.
Source: Hacker News · Launch HN · New tools have not been tested by long-term use. Evaluate them yourself.
Synthesia launched a plugin inside ChatGPT: describe a video or share a doc, deck or spreadsheet and it builds a finished video with avatars, voiceover, music and b-roll, then refines it with you in chat. Available on every Synthesia plan; search Synthesia in the ChatGPT Plugins menu and connect.
Source: Synthesia Blog · New tools have not been tested by long-term use. Evaluate them yourself.
An open-source (MIT) background scheduler for command-line agents: Claude Code, Codex and similar tools pick up and follow through on work from schedules or events, and only send you a card when they need a decision.
Source: GitHub · New tools have not been tested by long-term use. Evaluate them yourself.
The design platform Kittl now works inside ChatGPT and Claude: connect your account once, then ask for product shots, mockups, social content and video in the chat, using your Brand Kit.
Source: Kittl Blog · New tools have not been tested by long-term use. Evaluate them yourself.
Momentic (YC W24) took Mo, its AI QA engineer, out of beta: describe a flow in plain English and Mo tests it end to end against your web app or hosted iOS and Android devices, reporting bugs after a separate agent reproduces each one. Free to start with no credit card; benchmarks on its site are the vendor's own.
Source: Hacker News · Launch HN · New tools have not been tested by long-term use. Evaluate them yourself.
Canva's Affinity is now on iPad: sketch, vector, photo editing and page layout in one app and one file, built for Apple Pencil. Free for individuals (App Store download); whether its AI features, such as generative expand, upscaling and Decompose Layer, need a paid subscription: to be verified against the official page.
Source: 9to5Mac · New tools have not been tested by long-term use. Evaluate them yourself.
Tripo AI launched Agentic Canvas, an AI agent that builds 3D workflows on a visual node canvas: from model generation to texturing, retopology, rigging and export. Every step can be inspected and edited, and workflows can be saved and batch-reused, e.g. for game assets and 3D printing. Which plans include it: to be verified.
Source: PR Newswire · New tools have not been tested by long-term use. Evaluate them yourself.
An AI video editor its makers call "Cursor for video editing": ask for an edit in plain words and it changes the real timeline, press Tab to accept its suggested next cut, or edit the transcript like a document. Free plan; Windows, Mac, Linux and web.
Source: Hacker News · Show HN · New tools have not been tested by long-term use. Evaluate them yourself.
A Chinese-language beginner tutorial on AI agents that is also an agent you can run locally (Python): it starts from a 100-line minimal loop and adds one ability per lesson (files, commands, MCP, long-term memory), with a web view that shows every step.
Source: GitHub · New tools have not been tested by long-term use. Evaluate them yourself.
Liquid AI's demo Space that runs LFM3-Omni-2B locally in your browser on WebGPU; the small model understands text, images, video and audio and answers in text or speech.
Source: Hugging Face · New tools have not been tested by long-term use. Evaluate them yourself.
An open-source (Apache-2.0) AI assistant for Android phones: say what you need and it operates your apps and the web on a hidden background screen until the job is done, while you keep using your phone. Android 11+, installed as an APK.
Source: GitHub · New tools have not been tested by long-term use. Evaluate them yourself.
A free geology app for iPhone and Android: point your camera at the landscape and it overlays a geologic map in augmented reality, telling you what the rocks are and when and how they formed. Also works offline.
Source: Hacker News · Show HN · New tools have not been tested by long-term use. Evaluate them yourself.
An open-source (MIT) caption tool: transcribe and correct speech once, then cut captions for any screen shape (16:9, 9:16 and more) from one file; a CLI burns in and exports word-timed captions.
Source: Hacker News · Show HN · New tools have not been tested by long-term use. Evaluate them yourself.
An open-source AI assistant that runs locally on your device, positioned by its author as an alternative to Claude Cowork / ChatGPT.
Source: Hacker News · Show HN · New tools have not been tested by long-term use. Evaluate them yourself.
A terminal multiplexer for coding Agents that manages several Agent sessions at once.
Source: Hacker News · Show HN · New tools have not been tested by long-term use. Evaluate them yourself.
A small tool that shows in real time what each Claude Code / Codex Agent on your machine is doing.
Source: GitHub · New tools have not been tested by long-term use. Evaluate them yourself.
A programmatic video generation engine designed for Agents: generate videos with code.
Source: GitHub · New tools have not been tested by long-term use. Evaluate them yourself.
An open-source project positioned as an alternative to Claude Managed Agents.
Source: Hacker News · Show HN · New tools have not been tested by long-term use. Evaluate them yourself.
Black Forest Labs' demo Space for its FLUX image model on Hugging Face; useful for concept art.
Source: Hugging Face Trending · New tools have not been tested by long-term use. Evaluate them yourself.
The OpenMOSS team's demo Space for its image generation model.
Source: Hugging Face Trending · New tools have not been tested by long-term use. Evaluate them yourself.
Fish Audio's demo Space for its speech generation model; try it for character voices.
Source: Hugging Face Trending · New tools have not been tested by long-term use. Evaluate them yourself.
