Come join us in SF on Saturday to hack on Kontext with the BFL team
APPS
-
Video Team Enhances Anime Output Quality Standards
By
–
should "just work" – the video team really cares about anime output quality
-
Leverage AI Models as Specialized Tools for Better Results
By
–
TIP: Treat models like specialized tools—don’t rely on one model alone. Want deeper dives into AI tools, tips, trends, or comprehensive courses? Follow us @futurepedia_io and visit http://
Futurepedia.io to fully leverage AI in your work & life. -
GPT Compatibility Issues Between Reasoning Models o3 and 4o
By
–
Since reasoning models behave differently, any guidance on how this can impact GPTs built for non-reasoning models? So, if a creator intends for the GPT to function with 4o (which would be all GPTs prior to this change), but the user chooses o3…
-
Midjourney video launch, OpenAI scrutiny, and AI job trends
By
–
Top stories in AI today: – Midjourney's long-awaited video model
– AI watchdogs detail OpenAI concerns
– Automate browser tasks with natural language
– MIT study: ChatGPT’s detrimental cognitive impacts
– 4 new AI tools & 4 job opportunities Read more: https://
therundown.ai/p/midjourney-m
akes-its-video-debut
… -
MJ Video Animates Text Better Than Expected
By
–
MJ video does better at animating text than I expected.
— fofr (@fofrAI) 19 juin 2025
Here's an ideogram v3 image with text overlaying a photo, then the prompt:
> animate the letters of the text pic.twitter.com/uEP9THk6DAMJ video does better at animating text than I expected. Here's an ideogram v3 image with text overlaying a photo, then the prompt: > animate the letters of the text
-
Signal Refuses AI Integration to Protect User Privacy
By
–
'Meredith,' some guys ask, 'why won't you shove AI into Signal?' Because we love privacy, and we love you, and this shit is predictable and unacceptable. Use Signal
-
Hugging Face Launches Deepsite v2 with DeepSeek R1 Model
By
–
Hugging Face released Deepsite v2, the revamped version of its open-source vibe coding platform, with:
— Rowan Cheung (@rowancheung) 19 juin 2025
—Reasoning-capable DeepSeek R1 model
—Multiple inference options
—Incredible UI design outputs
—Targeted Edits on any site elementpic.twitter.com/ym7lMBc01kHugging Face released Deepsite v2, the revamped version of its open-source vibe coding platform, with: —Reasoning-capable DeepSeek R1 model
—Multiple inference options
—Incredible UI design outputs
—Targeted Edits on any site element -
Higgsfield AI Launches Canvas, Advanced SOTA Image Editing Model
By
–
Higgsfield AI released Higgsfield Canvas, a new SOTA image editing model It features advanced inpainting controls, allowing users to add products directly to an image or quickly change details of an output https://
x.com/higgsfield_ai/
status/1935042830520697152/video/1
… -
Midjourney Launches V1 Video Generation Model with Motion Controls
By
–
Midjourney dropped V1, its first-ever video generation model
— Rowan Cheung (@rowancheung) 19 juin 2025
An image-to-video system, V1 generates four 5s clips, which can be extended up to 20s — with motion/action controls
Priced at 8x image costs, which Midjourney says is 25x cheaper than rivalspic.twitter.com/zmRRcj8ZYwMidjourney dropped V1, its first-ever video generation model An image-to-video system, V1 generates four 5s clips, which can be extended up to 20s — with motion/action controls Priced at 8x image costs, which Midjourney says is 25x cheaper than rivals