yes, we will go from simply summarizing text to video and audio
INNOVATION
-
Downgrade from Version 3 to 2.5 Decision Made
By
–
We tried to use 3, but gave up and went back to 2.5.
-
Dense Models Like 2.5: Efficient LLMs for Better Performance
By
–
Dense models please, just like 2.5, only better and more up to date! 😀 We love 2.5 for FIM, but 3 is so expensive to use because it needs so much RAM :O
-

GPT-OSS-20B Scores 30% on AIME 2025 Math Exam
By
–
And the results from that eval, which runs 30 questions from the 2025 American Invitational Mathematics Examination 8 times each (240 prompts total) https://
static.simonwillison.net/static/2025/gp
t-oss-20b-aime25/gpt-oss-20b-low_temp1.0_20250816_094011.html
… -
Google’s Impressive Performance Over the Past Year
By
–
Google has been nothing but amazing in the last one year.
-
Unitree Wins 1500M Race at Robot Olympics
By
–
Unitree runs 1500M in 7:10, winning the Robot Olympics heat https://
search.app/aDNpv #WorldRobotGames #RobotAthlete #whrg #AI #ArtificialIntelligence #technology #tech #robotech #Robots #robotics #humanoidrobot #humanoid @fogle_shane @KanezaDiane @AlbertoEMachado -
World Humanoid Robot Games Opens in Beijing
By
–
World Humanoid Robot Games opens in Beijing! https://
search.app/YLrrp #whrg #AI #ArtificialIntelligence #technology #tech #robotech #Robots #robotics #humanoidrobot #humanoid @SpirosMargaris @PawlowskiMario @mvollmer1 @gvalan @ipfconline1 @LaurentAlaus @Shi4Tech @Fisher85M -
E-commerce Challenges: Events and Influencer Marketing Disruption
By
–
I’ve been in Ecommerce for 17 yrs and just lost on everything events and influencers these days!
-
Meta’s Open AI Infrastructure: Industry’s Hidden Messiness
By
–
AI infra is young, hard and messy (you obviously know this). the fact that meta talks about their infra details openly only makes it publicly messy. everyone's infra across the industry is pretty messy but privately so.
-
PyTorch Fork Infrastructure: Low Precision Types at Scale
By
–
from what i know (folks who worked on the infra bits at the other lab), the other lab forked pytorch several years ago for their large-scale runs before tensor subclasses were a thing, as they needed low and mixed precision types. now they have so much custom stuff on top of