Turn text into videos, images, and speech. Your face on an AI avatar? Done. Photos speaking? You got it! Ever wanted to clone yourself? Now, it's possible!
matrix
Joined January 2023
- Tweets4.7K
- Following16
- Followers1.6K
- Likes1.9K
The song was never the hard part. The edit was.
So we built the AI native editor.
Creatus Studio. Coming soon.
music.creatus.ai
music.creatus.ai × @alibaba_cloud
More compute. Newer models. Faster renders.
Quality goes up. Prices come down.
We just open sourced Memora, a memory architecture for AI agents.
Old way: new fact in, old fact erased, history gone.
Memora let you see what your AI agent believed on any date, and why.
100% accurate on "who said what, when" vs under 40% for standard memory.
Paper: doi.org/10.5281/zenodo.21654…
Code (MIT, open source): github.com/Nifty0x/memora
Creatus now has an MCP server 🥳
Connect once, then just chat with Claude, ChatGPT, Cursor ... Setup takes less than a minute.
CREATUS.AI retweeted
OpenAI CEO, Sam Altman:
«You don't need to write prompts anymore.»
In 38 minutes, he explains how to use ChatGPT better than most people do.
He gave this lecture for Stanford students. My friend sent me the recording.
I watched it last night, realizing I was using maybe 15% of what ChatGPT can actually do.
Watch it, then read the guide below on how to build a system that prompts itself.
We've seen crazy good outputs since our Producer launch 🤯
A user asked for Viking music video. One prompt.
2:38 video came back to a song that didn't exist 10min earlier.
One sentence in. No editor, no timeline, no second tool.
Sound on 🔊
music.creatus.ai
What if making a music video was just... a conversation?
Meet Producer. Tell it a vibe in one sentence. It writes your song, sings it, films the video with you in it, and puts it in front of an audience. All in one chat.
You type, it makes.
Avail at music.creatus.ai
JUST IN: Sakana AI research introduces Error Diffusion, training Dale-compliant dual-stream networks without weight transport. Achieves 96.7% MNIST & 61.7% CIFAR-10 accuracy. ED-PPO matches DFA-PPO in RL tasks. Key innovations enable task-dependent credit-assignment insights.
🆕 Creatus Community Radio
Creatus now has its own live radio. It plays real 100k+ tracks from the community around the clock.
Publish a track and it could be the next thing a new visitor hears. Your song could genuinely play next!
JUST IN: Linq introduces iMessage Apps bringing interactive experiences directly into conversations. Users can shop, play games, book flights, and make payments without leaving the iMessage thread. The API offers a seamless in-place update feature.
JUST IN: Anthropic releases Claude Sonnet 5, their most agentic model. Beats Sonnet 4.6 benchmarks, cheaper at $2/$10 intro pricing. Best for low/medium effort tasks. Sonnet 5 edges Opus 4.8 on some benchmarks. #AI #MachineLearning
JUST IN: Meta AI launched Brain2Qwerty v2, a non-invasive MEG brain-to-text pipeline. It decodes typed sentences with 61% word accuracy. The system reads MEG signals and reconstructs typed sentences in real time without implants or surgery.
JUST IN: EverMind releases EverOS, an open-source memory runtime for AI agents. It stores memory as Markdown files, uses hybrid BM25 + vector retrieval, and generates self-evolving Skills. EverOS is local-first, compatible with OpenAI, and offers cloud hosting options.