multimodal models
-
Liquid AI deploys LFM models for BMW and Roche
Liquid AI has released Liquid Foundation Models designed for sequential multimodal data, including a 40.3B Mixture of Experts version. These models support…
-
Yann LeCun’s Meta legacy and the shift to world models
Yann LeCun’s leadership at Meta drove the development of PyTorch, which holds a 26.34% market share. His departure to launch Advanced Machine…
-
Adept’s Fuyu-Heavy and the Salesforce Agentforce pivot
Adept’s Fuyu-Heavy model outperforms Gemini Pro on multimodal reasoning benchmarks by using pixel-based screen perception. Meanwhile, Salesforce transitions toward autonomous agents via…
-
Stability AI management and Stable Diffusion 3.5 vs Midjourney
Stability AI manages capital from a $101 million raise while navigating competition with Midjourney. While Midjourney offers superior lighting and composition, Stable…
-
Runway’s Act-One changes how creators animate digital characters
Runway’s Act-One tool replaces expensive motion capture hardware by mapping actor performances onto AI characters using a single camera. This technology allows…
-
Hugging Face scaling and SmolVLM edge multimodal deployment
Hugging Face maintains a $4.5 billion valuation while its SmolVLM models achieve high efficiency, with the 256M variant requiring less than one…
-
Deploying Llama 3.2 90B Vision on SambaNova SN40L
SambaNova’s SN40L RDU provides enterprise infrastructure for Llama 3.2 90B Vision, delivering a 3.7x speedup over the DGX H100. The hardware uses…
-
Runway Gen-4.5 leads video benchmarks as Sora faces discontinuation
Runway’s Gen-4.5 model leads the Artificial Analysis Text to Video benchmark with 1,247 Elo points, outperforming Google and OpenAI. The platform offers…
-
The collapse of Disney’s OpenAI deal and the rise of Kling AI
Disney’s three-year licensing deal with OpenAI collapsed after the company deprecated Sora to preserve compute. Meanwhile, ByteDance’s Seedance 2.0 and Kuaishou’s Kling…
-
Evolutionary model merging and Sakana AI’s expansion
Sakana AI reached a $2.65 billion valuation using evolutionary algorithms to merge generative models like EvoLLM-JP, which outperforms Llama 2 70B and…
-
Moonshot AI targets $50 billion valuation amid model transitions
Moonshot AI seeks a $50 billion valuation as it transitions from legacy K2.5 models to the K3 multimodal flagship. The new K3…
-
Deploying Llama 3.2 90B Vision on Groq LPU for Real-Time Performance
Groq’s LPU architecture delivers 750 to 800 tokens per second for Llama 3 70B, significantly outperforming NVIDIA H100 clusters by eliminating memory-bandwidth…