Sabitlenmiş Tweet
Ideogram
1.3K posts
Ideogram
@ideogram_ai
Think it. Make it. Own it. Research and product company building open frontier models for generative media and design.
Katılım Mayıs 2023
1 Takip Edilen71.4K Takipçiler
First part of our series on how we built one of the fastest image generation systems, and we're kicking off with running LLMs (for prompt expansion) at >1000 tok/s by leveraging a DSpark adapter we trained from scratch!
blog.fal.ai/how-we-achieve…
English
@sgl_project @fal Hi 👋 @sgl_project -- we also use your framework to serve our magic prompt language model similar to our friend @fal 😍
English
The first production DSpark deployment, running on SGLang ⚡️
@fal benchmarked and picked SGLang for faster low-batch, high-interactivity serving. The result: 16x throughput on a single B200, sub-2s prompt expansion for Ideogram V4
Huge congrats to the team 👏
Doğaç@dogacel0
At @fal we adapted DSpark to our serving stack and got 16x throughput over the baseline. We think this is the first production deployment of DSpark since DeepSeek V4, running on SGLang ⚡️
English
AI Poll: What's your favorite image model currently?
Overall, for art, infographics, and perhaps for other use cases?
English
these beautiful images were rendered locally on a MacBook using an open weight model.
@ideogram_ai you have built a great product 🤌🏼
English
Today felt different 🥹🙏
I'm just a creator posting AI images every day, so seeing @ideogram_ai notice my work genuinely meant a lot.
Big thank you to the team for the recognition ❤️
Moments like this remind creators that their work is being seen, We'll keep creating.
English
@javi_22_dev @Kon237530842750 We're working on image editing model too 🤫
English
Today we published a technical blog post about Ideogram 4.0 — our goal is to enable more innovation and creativity.
It's a 9.3B Diffusion Transformer trained from scratch, paired with a frozen 8B VLM as text encoder. The nf4 checkpoint runs on a 24GB consumer GPU.
Thread 🧵
English
After testing of ideogram4, it is a SOTA model.
I quickly find that JSON captions + bbox works very well, especially in terms of precise positioning.
I think they need to make more breakthroughs:
Since BBOX is being used, why not further split layers with transparent layer?
Justine Moore@venturetwins
Ideogram just released the #1 open weight image model in the world. It punches way above its weight - it's small enough to run on a consumer GPU, but goes head-to-head with Nano Banana and GPT Image for design. We went deep with @mo_norouzi on how they did it + hot takes 👇
English
Ideogram retweetledi
Going live with the founders of @ideogram_ai and ComfyUI.
@mo_norouzi (CEO, Ideogram) + @yoland_yan (CEO, ComfyUI ) sit down with @PurzBeats & @hellorob to talk building open-weight text-to-image models, where open weights stand vs. closed APIs, and what's next.
What we'll cover:
→ Founders interview
→ Ideogram 4.0 + open-weight release
→ Structured JSON prompting & bounding box layout control
→ Live community Q&A Drop your questions in chat
we'll put them to the founders live.
English
@venturetwins @mo_norouzi Excited to keep shipping open frontier models for design and creativity 🚀
English
Ideogram just released the #1 open weight image model in the world.
It punches way above its weight - it's small enough to run on a consumer GPU, but goes head-to-head with Nano Banana and GPT Image for design.
We went deep with @mo_norouzi on how they did it + hot takes 👇
English
Ideogram retweetledi
So proud of what the @ideogram_ai team pulled off. The #1 open weight image model in the world and small enough to run on your consumer GPU 🚀
Justine Moore@venturetwins
Ideogram just released the #1 open weight image model in the world. It punches way above its weight - it's small enough to run on a consumer GPU, but goes head-to-head with Nano Banana and GPT Image for design. We went deep with @mo_norouzi on how they did it + hot takes 👇
English
Ideogram v4 is now in @pixio_ai 🚀
Open source. API-powered. JSON Prompt-ready.
Creators can now control visual placement, composition, text, and style with way more precision — directly through Pixio.
Huge respect to @ideogram_ai for opening this up to the community.
English
@ideogram_ai thank you for featuring the Arcane style lora, lovely model to work with, addicted to boxes.
Looking forward to editing with references and boxes in the future .. wink wink
English
One week ago, we released Ideogram 4.0, our first open weight model.
Since the launch, the open-source and creative communities have pushed it beyond what we could imagine. Seems like JSON and bounding box prompting is becoming the new norm now.
Week one community highlights (thread):
English
looking forward to when this kind of conditioning makes its way into 3D generators. Such a great way to give control back to the artist.
Ideogram@ideogram_ai
One week ago, we released Ideogram 4.0, our first open weight model. Since the launch, the open-source and creative communities have pushed it beyond what we could imagine. Seems like JSON and bounding box prompting is becoming the new norm now. Week one community highlights (thread):
English
@ideogram_ai You really released something great here. Thank you.
English