Syntographer, Engineer, Optimist, Ideologist Interested in #MachineLearning,#AI #AiArtwork #AIart #aiartcommunity
Красноярск, Россия
Joined February 2011
- Tweets14.4K
- Following541
- Followers1.5K
- Likes1.4K
Sergey Bogorad retweeted
I sent the following message to our team and investors:
—
As you know, Daniel Gross’s time with us has been winding down, and as of June 29 he is officially no longer a part of SSI. We are grateful for his early contributions to the company and wish him well in his next endeavor.
I am now formally CEO of SSI, and Daniel Levy is President. The technical team continues to report to me.
You might have heard rumors of companies looking to acquire us. We are flattered by their attention but are focused on seeing our work through.
We have the compute, we have the team, and we know what to do. Together we will keep building safe superintelligence.
Ilya
Sergey Bogorad retweeted
This SREF is so good. Hand-painted children's illustrations. I'm still not quite sure how to prompt away the sheet of paper. Any ideas?
--sref 3854775940 --sv 6
#midjourney #sref #aiart #tischeins
Sergey Bogorad retweeted
35mm Film = Video ⚡️
Ilford 35mm black and white wide-angle photo of Bone Tomahawk, motion blur, golden hour, asymmetrical, cinematic, film contrasts, expired film aesthetics, film grain and dust, depth of field --ar 3:2 --exp 25 --raw --s 75 --v 7
Video images in comment ⚡️
Sergey Bogorad retweeted
1. Midjourney enters the video game with their first AI video model V1 that turns any image into 5-second clips.
Users can also extend videos up to 20s and it's available on their $10/month plan starting now.
Sergey Bogorad retweeted
So good! ❤️ Let’s see your animation, please!
Illustration [prompt] --ar 2:3 --exp 42 --sref 3711281041 --profile fbichjd --v 7
#midjourneyv7 #srefcode
Sergey Bogorad retweeted
Tons of crazy releases from Google I/O 🤯
- Agent Mode
- Google Veo 3
- Jules Code Assistant
- AI Filmmaking Tool Flow
- Project Astra live abilities
- AI Image Google Imagen 4
- Think mode & native audio
- Gemini 2.5 new capabilities
Here’s EVERYTHING you need to know:
Sergey Bogorad retweeted
4 SREF codes that work really well for me with v7. All the prompts are in the ALT. I’m curious to see if you like them.
--sref 1309180883
--sref 648811750
--sref 3462541816
--sref 2170633035
#midjourney #sref #aiart #tischeins
ALT Prompt: cinematic shot, hyper-realistic photograph of a half-elf female barbarian, muscular but graceful build, tribal leather and fur armor. Wild dark hair with braided sides, fierce expression with elven features. Battle weapon strapped to back. Tribal paint on face. Combat-ready pose. Dramatic lighting, detailed armor texture. --sref 1309180883 --ar 5:4 --v 7.0 --sv 4
ALT PROMPT: cinematic shot, hyper-realistic photograph of a half-elf female barbarian, muscular but graceful build, tribal leather and fur armor. Wild dark hair with braided sides, fierce expression with elven features. Battle weapon strapped to back. Tribal paint on face. Combat-ready pose. Dramatic lighting, detailed armor texture. --sref 648811750 --ar 5:4 --v 7.0 --sv 4
ALT PROMPT: A model like figure, girl, fighter, big, young, long braided black hair, training in a puddle of water in front of a waterfall, sword, scar --chaos 30 --ar 5:4 --stylize 300 --sref 3462541816 --v 7.0 --sv 4
ALT PROMPT: A model like figure, girl, fighter, big, young, long braided black hair, training in a puddle of water in front of a waterfall, sword, scar --chaos 30 --ar 5:4 --stylize 300 --sref 2170633035 --v 7.0 --sv 4
Sergey Bogorad retweeted
Future Tools Update: In the spirit of building in public, I wanted to share that I am still single-handedly "vibe-coding" the new version of the site. It's taking a bit longer than I anticipated but, it's not because the coding isn't working. It's because other priorities keep getting in the way. I am making a little bit of progress every day though.
These new models like Gemini 2.5 and OpenAI's ImageGen are actually making it easier to pick up the pace with less issues. I also just learned that Perplexity has an MCP (thanks @rileybrown_ai), making it easier to troubleshoot issues directly from within my IDE (usually WindSurf, sometimes Cursor).
As of now, I've completely cloned the existing functionality of the current site with a new design. Filtering, sorting, and searching tools currently works perfectly. The back-end where I manage the tools is working as well. I've even already built automations in Make dot com that automatically update the news page in the same way the current site works.
I've added the ability for users to login to the site in order to favorite tools and view/filter by their favorites. I've built out a "Gadgets" page that works just like the existing tools page but for physical AI gadgets. I've built out an events page to keep people looped in on upcoming AI and tech events. I've overhauled the "Videos" page to better sort and filter to specific video types. Things are coming along nicely.
Functionality that I'm still building...
The individual gadgets pages still need to be dialed in to best show off each gadget. The events page is kind of ugly and needs a design upgrade (but it works). I'm adding the ability for users to comment/review products if they're logged in so the community can share thoughts on tools. I'm building a leaderboard so that the most active community members can earn rewards for reviewing tools and helping us find dead tools or scams. The "Featured Listing" functionality where sponsors can boost tools hasn't been implemented yet. I want to add a natural language chat option so people can find tools by explaining the problems they're trying to overcome. I also need to make sure the SEO is dialed in so that, when I migrate to the new site, I don't lose all my existing rankings. Finally, I need to do a ton of testing to make sure the site can handle a lot of traffic and is safe against hackers and malicious bots.
I've got big plans for this site and I may not have them all ready when I launch but I'm getting a little closer every day. I imagine I won't have the leaderboard and the natural language chat functionality ready when I launch but pretty much everything else should be good to go. The other features will follow soon after.
I'm also debating about rebranding with a new logo. I've had a few people say they're not fans of the logo. I like it because it was part of a contest and it was created by a fan of the site and voted on by community members. It's also got subtle / hidden "FT" text in it. However, it also looks a bit like the OpenAI logo and a few other tech brand's logos. Maybe I'll run another contest and give the winner something really cool. I could use AI to generate some ideas but I like getting the community involved in things like this... Open to feedback / ideas here.
Anyway... That's the update. I'm super excited about getting this live. I think it's a big improvement. But I'm always open to ideas and feedback as well.
Sergey Bogorad retweeted
Ok. Imagegen from OpenAI does look pretty good. Excited to get my hands on it and test it side-by-side with Google's native Image generation.
Sidenote, what's with OpenAI and Google just using the same name for products lately? Imagen Vs. Imagegen, Canvas, Deep Research...
Sergey Bogorad retweeted
Excited to come out of stealth at @reveimage!
Today's text-to-image/video models, in contrast to LLMs, lack logic. Images seem plausible initially but fall apart under scrutiny: painting techniques don't match, props don't carry meaning, and compositions lack intention. (1/4)
The Halfmoon 🌓 reveal: Congratulations to @reveimage on creating the world’s leading image generation model with Reve Image!
Reve Image has been in the Artificial Analysis Image Arena over the past week and is the clear leader, beating strong competition including Recraft V3, Google's Imagen 3, Midjourney v6.1, and Black Forest Lab's FLUX.1.1 [pro].
The model stands out for its impressive text rendering, prompt adherence, and aesthetics. Currently, the model is accessible through Reve’s website (link below), an API is not yet available.
See below for example images and a link to try the model out for yourself 🧵
Sergey Bogorad retweeted
Super excited to share 🧠MLGym 🦾 – the first Gym environment for AI Research Agents 🤖🔬
We introduce MLGym and MLGym-Bench, a new framework and benchmark for evaluating and developing LLM agents on AI research tasks.
The key contributions of our work are:
🕹️ Enables the exploration of different training algorithms for AI Research Agents such as RL
🛠️ Provides a flexible evaluation framework that can accommodate different artifacts such as models, algorithms, or predictions
🤖 Allows researchers to evaluate any model without the need to develop a custom agentic harness
🎯 Introduces 13 diverse open-ended AI Research tasks for evaluating AI Research Agents on a wide range of domains such as computer vision, natural language processing, reinforcement learning, game theory, and logical reasoning.
📈 Proposes a new evaluation metric for AI Research Agents
MLGym makes it easy to:
1) Add new tasks
2) Evaluate new models
3) Integrate new agents
Check out a video of the MLGym Agent to see how it performs the full pipeline of idea generation💡, implementation 👩💻, experimentation 👩🔬, and iteration 🔄 to improve on ML tasks.
Huge thanks to the exceptionally talented @deepaknathani11 who led this work and to all the other amazing collaborators who made this possible 🙏🫶🚀
Sergey Bogorad retweeted
I found an amazing girl! She sings rhythmically by connecting Japanese tongue twisters!🇯🇵😳
I’ll give it a try!😎🎙️
Sergey Bogorad retweeted
Wait…WHAT? Introducing Pikadditions, the easiest way to make your content stand out.
Add anyone or anything to any video, whether that’s a video you shoot yourself, or a favorite clip. Special surprise: get fifteen free Pikadditions generations when you sign up!
Go try it at pika dot art
Sergey Bogorad retweeted
Alright, here's what I've rounded up for AI news between last Friday and today...
- OpenAI unveils o3-mini model for cost-effective reasoning
- OpenAI introduces 'deep research' tool for complex tasks
- OpenAI adds web search to o3-mini in early prototype stage
- OpenAI makes ChatGPT search public without sign-up
- Free ChatGPT search & reasoning now powered by o3-mini
- OpenAI raises ChatGPT memory limit by 25% for Premium
- Sam Altman admits OpenAI erred on open source stance
- Figure AI quits OpenAI deal, teases humanoid robotics leap
- Gemini 2.0 arrives with Flash, Flash-Lite, and Pro variants
- Google debuts Imagen 3 image tech via Gemini API
- Google drops weapons ban from its AI principles
- Amazon readies major AI upgrade for Alexa assistant
- Lyft taps Anthropic’s Claude AI for customer support
- Anthropic challenges users to jailbreak its new AI
- Mistral AI launches Le Chat, a new assistant for web/mobile
- GitHub Copilot introduces "Agent Mode"
- Pika debuts Pikadditions for AI-generated video overlays
- PikaScenes transforms pet photos into fantasy worlds
- Topaz Labs unveils AI-based video restoration 'Starlight'
- Snap reveals mobile AI text-to-image model for creators
- X adds Aurora model for AI-driven image edits in Grok
- Beatles bag Grammy for AI-enhanced track 'Now and Then'
- EU bans AI with ‘unacceptable risk’ in sweeping crackdown
- US bill could jail DeepSeek downloaders for up to 20 years
- Karpathy releases 3-hour LLM deep-dive on ChatGPT tech
A TON seems to be happening every week. I'm getting ready to record my breakdown of it all for tomorrow's video. Is there anything big that I'm missing that I need to include?
Sergey Bogorad retweeted
ChatGPT search is now available to everyone on chatgpt.com — no sign up required.
