AI Primer
Your daily AI changelog: latest releases, frameworks, and infrastructure updates for AI engineers.
www.ai-primer.comANIMATION AND VISUAL EFFECTS
MiniMax H3 Max workflow uses staged prompts to control timing
Glenn Has A Beard found that MiniMax H3 Max can miss object orientation and delayed cause-and-effect. His tests lock the camera, specify a final state, and stage timing constraints one at a time.
COMPUTER VISION AND PATTERN RECOGNITION
OpenArt Arena ranks image and video models by production task with blind pairwise judgments
OpenArt Arena uses blind pairwise judgments from creative professionals to rank image and video models for ads, film, animation, editing, and lip sync. It also publishes overall leaderboards.
GENERATIVE AI AND FOUNDATION MODELS
Google releases Gemini 3.8 Live with visual understanding
Google introduced Gemini 3.8 Live and Live Extended Thinking in Gemini Live and its API. The models add near-real-time visual understanding, support for 97 languages, and background tool use.
GOOGLE DEEP MINDGOOGLE
MOBILE APP DEVELOPMENT
Devin adds cloud Mac sandboxes for Xcode and iOS app testing
Cognition says Devin can launch native macOS environments, run Xcode and iOS simulators, and test Apple apps in the cloud. The company demonstrated prompt-driven workflows for iOS, iPadOS, and macOS apps.
DATABASES AND DATA ENGINEERING
Perplexity reports coding agents helped build CobbleDB, its DynamoDB replacement
Perplexity says two engineers and hundreds of persistent coding agents built CobbleDB, an internal key-value database for its search stack. It is optimized for repeated batch reads of prepared page records and is not offered externally.
