grok-imagine-video-1.5 now handles text-to-video, image-to-video, and reference-to-video with up to seven image inputs, three audio inputs for voice consistency, and native 1080p output, rolling out across Grok, iOS, Android, and xAI API for paid subscribers.
Why it matters: AI video generation cleared another capability threshold at $4.20/min vs Sora's $30 — the cost gap makes Grok the default to evaluate first for short-form creative workflows.
Korea's largest open-source LLM uses MoE architecture (37B active params), supports 10 languages and a 262K-token context window, and ships under Apache 2.0 for unrestricted commercial use.
Why it matters: Another frontier-scale open-weights model with permissive commercial licensing expands the options for cost-sensitive builds where API pricing is a constraint.
Instead of launching a standalone Android/iOS app, Google is integrating AI Studio's build-and-prototype capabilities directly into the Gemini app on mobile and desktop; the web-based aistudio.google.com stays live for developers.
Why it matters: Google is consolidating developer tooling around Gemini as the single interface — teams building on Google AI should plan for Gemini, not a separate app, as the environment going forward.