Linker Vision is transforming city operations in Kaohsiung city with NVIDIA AI. Using city-scale #digitaltwins and video analytics #AIagents built with @NVIDIAOmniverse and #NVIDIAMetropolis, they’re optimizing urban ops and reducing response times to incidents by up to 80%.
MULTIMODAL AI
-
AI Model Demonstrates Impressive Accents and Situational Awareness
By
–
The accents in these are fantastic. @altryne also pointed out the really incredible situational awareness going on. https://t.co/q86w3HQ2UA
— fofr (@fofrAI) 22 mai 2025The accents in these are fantastic. @altryne also pointed out the really incredible situational awareness going on.
-
Gemini 2.5 Pro Deep Think Outperforms on Complex Reasoning Benchmarks
By
–
Early tests of Gemini 2.5 Pro Deep Think are looking 🔥
— Google AI (@GoogleAI) 22 mai 2025
By applying new research techniques in reasoning, Deep Think is outperforming leading general foundation models on complex reasoning benchmarks, including the challenging USA Math Olympiad benchmark & MMMU. pic.twitter.com/SkM5jn7Q4VEarly tests of Gemini 2.5 Pro Deep Think are looking By applying new research techniques in reasoning, Deep Think is outperforming leading general foundation models on complex reasoning benchmarks, including the challenging USA Math Olympiad benchmark & MMMU.
-
AI Video Generation Encoding Bug: Mismatched Lip Sync
By
–
No idea, could just be an encoding bug or something. It's odd because I can see the lips move as though they are saying the words I asked for. (The woman on the right says "foffer") pic.twitter.com/AFNWZ1CPaJ
— fofr (@fofrAI) 22 mai 2025No idea, could just be an encoding bug or something. It's odd because I can see the lips move as though they are saying the words I asked for. (The woman on the right says "foffer")
-
Mistral Enhances OCR with New Features
By
–
Mistral AI has released an update to the Mistral OCR! – New OCR model with overall improvements
– BBox Annotations: Structured outputs for extracted images
– Document Annotations: Structured outputs for documents -
NVIDIA Cosmos Wins Best Choice and Golden Award at COMPUTEX2025
By
–
We're thrilled to announce that #NVIDIACosmos has won both the Best Choice of the Year & the Golden Award at #COMPUTEX2025. Read how Cosmos is built to accelerate synthetic data generation and push the boundaries of #PhysicalAI for autonomous machines.
-
Runway References: Image-Based Prompting for Consistent Characters
By
–
introducing Runway References.
— KREA AI (@krea_ai) 22 mai 2025
this new model lets you reference images on your prompts to keep characters or scenes consistent.
powered by @runwayml pic.twitter.com/D5aPp83H0rintroducing Runway References. this new model lets you reference images on your prompts to keep characters or scenes consistent. powered by @runwayml
-
Subtitle Generation Accuracy Issues Impact AI Content Quality
By
–
It's amazing how much the inaccuracy of the generated subtitles can ruin a generation.
-
Flow continuity engine improves scene transitions in Dear Stranger
By
–
The scene transitions in Dear Stranger are shockingly smooth for AI. Flow’s continuity engine is doing something right.