Topic: multimodal ai models
-
Nvidia Nemotron 3 Nano Omni: 30B params, 3B active, for edge AI
Nvidia launched the open-weight Nemotron 3 Nano Omni, a multimodal AI model with 30 billion total parameters but only 3 billion activated per inference via a mixture-of-experts design, enabling real-time vision, audio, and language processing on a single GPU for edge deployment. The model deliver...
Read More » -
DeepSeek debuts experimental multimodal model to challenge Anthropic
DeepSeek released DeepSeek-V4-Flash-Vision-Exp, a multimodal agent model claiming performance close to Anthropic's Opus-4.8, but its own benchmarks show it wins on only 3 of 11 evaluations, with notable deficits (e.g., 12 points on NL2Repo). The vision model's major improvement over the text-only...
Read More » -
Google Cuts Top-Tier AI Plan Price as Gemini Gains Power
Google is reducing the price of its premium AI Ultra plan from $250 to $200 per month and introducing a new $100 tier, announced at Google I/O. A new AI assistant called Gemini Spark will integrate with Chrome and a new Android interface called Android Halo to help users complete tasks across app...
Read More » -
Meta’s AI Agent Muse: New Features & Updates
Meta is positioning its new AI agent, Muse, as a central hub for productivity and lifestyle management through integration with existing digital ecosystems and upcoming hardware like smart glasses. The company plans to monetize the free service by charging transaction fees on purchases facilitate...
Read More » -
Samsung, LG Uplus to trial 6G sensing as radar alternative
Samsung Electronics and LG Uplus have signed an MOU to jointly develop Integrated Sensing and Communication (ISAC) technology, which repurposes cellular base stations to function as environmental sensors by analyzing wireless signals bouncing off objects. The International Telecommunication Union...
Read More » -
Pichai Admits Google Is Behind on AI Coding
Google CEO Sundar Pichai admitted the company is "a bit behind" in agentic coding, tool use, and long-horizon tasks, acknowledging a gap compared to competitors despite Google's strengths in other AI areas. Pichai attributed the lag to a developer product gap, noting Google lacked an external cod...
Read More » -
Google Maps Adds AI Captions with Gemini
Google Maps is using its Gemini AI to automatically suggest captions for user-shared photos and videos, starting with iOS users in the U.S. and planning a global Android rollout soon. The feature aims to increase valuable user contributions by reducing the hesitation to write captions, thereby en...
Read More » -
Google upgrades Asset Studio with Gemini video and creative tools
Google announced at Marketing Live 2026 that Asset Studio is getting Gemini-powered upgrades to generate text, images, and video from natural language prompts, including new video creation workflows via Gemini Omni. A new 1-Click Creative Testing feature automatically identifies high-performing a...
Read More » -
4 reasons the Gemini-ChatGPT gap is rapidly closing
Google Gemini has rapidly closed the gap with ChatGPT, growing from 90-140 million monthly users in 2024 to 750 million by 2026, thanks to faster growth and OpenAI's PR challenges including a rocky GPT-5 launch and a crime planning incident. Gemini offers superior value through bundled services, ...
Read More »