Sunday Rundown #70: Flux1.1 Pro & Pinhead's Baby
Sunday Bonus #30: How to get FLUX1.1 [pro] for free.
Happy Sunday, friends!
Welcome back to the weekly look at generative AI that covers the following:
Sunday Rundown (free): this week’s AI news + a fun AI fail.
Sunday Bonus (paid): a goodie for my paid subscribers.
Let’s get to it.
🗞️ AI news
Here are this week’s AI developments.
👩💻 AI releases
New stuff you can try right now:
Black Forest Labs released FLUX1.1 [pro], the latest version of its image model. It is cheaper, better, and faster than its predecessor FLUX.1 [pro] (as well as most other text-to-image models.)
ChatGPT is getting a beta feature called Canvas, which looks suspiciously similar to Claude Artifacts. Canvas brings up a separate work window to the right of your chat, which displays your WIP code, documents, and so on.
Google had a busy news week:
Google Lens can now interpret live video and voice input to give you on-the-spot answers about the world around you.
The new Gemini 1.5 Flash-8B almost matches the performance of its bigger Gemini 1.5 Flash cousin at half the cost and twice the rate limit.
The overhauled “Summary cards” in Gmail will have better visuals along with an improved way to surface and display important information.
Luma Labs made Dream Machine video generation 10x faster.
Microsoft is giving its Copilot a major upgrade, including new voice and vision capabilities, Copilot Labs that lets you test beta features, and more.
NVIDIA released an open-source model called NVLM-D-1.0-72B (so catchy) that’s on par with most frontier LLMs.
OpenAI held its DevDay where it announced new ways for developers to build apps with its Advanced Voice Mode, improved vision capabilities, and more.
Pika Labs launched version 1.5 with more realistic movements, better camera controls, and cool “Pikaeffects” that let you e.g. crush and explode objects. (The downside is that free users no longer have access to the classic Pika 1.0 model.)
Pinterest has new AI and automation features for advertisers that let them optimize their ads and generate product backgrounds.
🔬 AI research
Cool stuff you might get to try one day:
ByteDance showcased two impressive video models—PixelDance and Seaweed—which are in invite-only beta and aren’t widely available yet.
Meta announced a new text-to-video model called Meta Move Gen. It lets you create and edit videos using text as well as generate sound effects and soundtracks.
Microsoft will roll out new AI features to Copilot+ PCs in November, including “Recall,” improved Windows search, new generative tools in Photos and Paint, and more.
📖 AI resources
Helpful stuff that teaches you about AI:
What Are AI Agents—And Who Profits From Them? - a good dive into the agent landscape by Every.
🔀 AI random
Other notable AI stories of the week:
Google is said to be working on its own “reasoning model” like OpenAI’s o1.
OpenAI raised a whopping $6.6 billion at a $157 billion valuation.
🤦♂️ AI fail of the week
“Even monsters deserve a family.” - DALL-E 3, unprompted. (Final version.)
💰 Sunday Bonus #30: These two sites actually let you use FLUX1.1 [pro] for free
FLUX1.1 [pro] is all the rage this week.
Much like FLUX.1 was the week I pitted it against other image models.
FLUX1.1 [pro] is officially available here:
The thing is, you’ll have to pay to use any of these sites.
(Although to be fair, at 4 cents/image, it’s not exactly a life-ruining amount of money.)
Other sites provide free image credits and list the FLUX1.1 [pro] model in their library:
However, they come with major caveats:
Flux1.ai gives you 10 free credits when you sign up, but one FLUX1.1 [pro] image costs 8 credits. I
wastedwisely used my credits on that beautiful robot artist above.Freepik and BasedLabs don’t give you enough free credits for even a single FLUX1.1 [pro] run.
So they are “free” only in theory.
But there are two other sites that actually let you use FLUX1.1 [pro] for free.
Let me show you how to use them.