not much to say. pace has been incredible. 4 rounds in 18 months. incredible new investors @sonyatweetybird @mamoonha. all metrics going 🆙 exponentially. new exclusive models. more capital = more GPUs = more people = more investment into gen media. back to work.

I built loratlas.com to make it easier to browse the 1,500 Krea 2 Style LoRAs I trained. You can try them with your own prompts, run a random loop to generate lots of examples, or blend multiple styles together. Images can be generated directly from the site using @fal API keys, or you can grab the LoRA weights from the @huggingface link on each style page. Every LoRA was trained for just 100 steps. If you like one, you can always build a larger dataset around it and retrain it with more steps.

latest adventures at @fal design lab: Türkiye x USA soccer scarves

Deploying megawatts. Gigawatts 🔜🔜🔜

Work by amazing researchers and proud to contribute to this effort with GPUs from our Generative Media Cloud 👇👇👇

Check out our new work led by brilliant @AbbasMammadov11 and @ozgurkara99

Looking for up to 10 MW available in the next 120 days. USA. LC is preferred but can do with Air.

yimby but for datacenters

infallible

If you are running inference on your own models, come talk to us. Already serving primary inference needs for frontier image and video model labs. Investing very heavily into fal.ai/serverless.

Personally hiring Distributed Systems Engineers in San Francisco. Rust and Python. One of the fastest growing companies, to date. With the leader in the vertical.

😌😌😌😌

biggest gen media event of the summer

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 👇

Looking for >5MW space anywhere in north america (ideally US). Available in the next 120 days. DM me

ML perf is not only about writing faster kernels but also about speeding up disk to GPU transfer and improving setup times. i love how FlashPack (made by fal) aims to solve the former. should i write an article going deeper into how it works?

ok let’s recap my first week at @fal 1. quit smoking 2. gym streak is still alive, cardio sessions hit hard 3. started cooking at home. not just boy dinner, actual salads and greens 4. still shocked by how smart everyone is here 5. they ship fast. i mean pretty damn fast 6. culture is really good 7. got recognized at a few cafes because of the fal sticker on my laptop. still weird because i didn’t know this many random people in Türkiye knew fal 8. Batuhan, Gorkem and Burkay are very included in meetings, which surprised me. idk, billion dollar company brain made me think it would be harder to see them this often 9. as a new hire, i thought i wouldn’t be taken that seriously at first, but i feel very included. also special thanks to Ilker and Gokay. such a pleasure to work with them so yeah, definitely recommend applying to fal

Today we launched Reve 2.0! We built it from the ground up with a completely new architecture. All visual generative models today use text as intermediate representation, leveraging Large Language Models to plan their outputs before rendering any pixels. Natural language is expressive, but it is ambiguous. Ambiguity is the enemy of control. Two years ago, we made a different bet. We replaced text with a better, code-like semantic representation — a layout. Layout is the reason we can compete with models trained on 10× our compute. It opens up a whole new world of precise, non-verbal visual control. Very proud of the Reve team for this incredible achievement!

Download weights here: github.com/ideogram-oss/ideo… Or try it on Fal: fal.ai/models/ideogram/v4

iykyk