AI News: Claude Opus 4.8, Insane Omni Use-Case, and A Dog Translator?

COClaude Opus 4.8AnthropicAvailable in EU

Minor upgrade over Opus 4.7 with slightly better coding, reasoning and computer use, and improved honesty (flags uncertainty, fewer unsupported claims). Available on all platforms at unchanged pricing.

DWDynamic WorkflowsAnthropicAvailable in EU

Claude Code feature that plans dynamically, splits a prompt into subtasks, fans them out to parallel subagents, and has agents cross-check and refute each other until answers converge. The creator hasn't tested it yet.

MIMAI Image 2.5MicrosoftUnverified

Microsoft's image model ranks third on the Arena leaderboard behind GPT Image 2 and Gemini 3.1 Flash. Better instruction following and text rendering; the creator tested a flyer prompt on arena.ai with a solid result.

M3Microsoft 365 Copilot redesignMicrosoftAvailable in EU

New Copilot design with a longer prompt box, inline formatting, and pulling data from emails, files, chats and meetings, with inline charts. Available across Excel, PowerPoint and Word.

PCPerplexity Computer (local on Mac)PerplexityUnverified

Perplexity Computer is now available inside Word, Excel, PowerPoint and Outlook for complex multi-step tasks, such as drafting negotiation redlines with track changes.

LILeonardo Image to 3DLeonardo AIUnverified

Leonardo added image-to-3D generation with a 3D reference view creator that produces multiple angles for better detail. Tested with a wolf and the creator's own photo; takes about 5 minutes, and results were decent but a bit cursed on faces.

EMElevenLabs Music v2ElevenLabsAvailable in EU

Updated music model trained on licensed data and cleared for commercial use. The creator generated a pop-punk song about tacos in San Diego, which included Padres references.

EDElevenLabs Dubbing v2ElevenLabsAvailable in EU

Dubbing feature that keeps the speaker's voice, emotion and expressions. Free tier dubs up to 30 minutes; the creator dubbed 10 minutes of a video into Hindi but couldn't judge quality.

GOGemini OmniGoogleUnverified

Google's Gemini Omni model released last week. Use cases shown: generating a first-person taxi drive from a route drawn on a map screenshot, and drone POV footage following a sketched camera path.

SASpotify AI Persona BadgeSpotifyUnverified

YouTube moves AI disclosure labels to a more prominent position and adds automatic detection that labels significant photorealistic AI content when creators don't disclose.

MJMeng Jiao Yi AI Pet TranslatorMeng Jiao YiUnverified

A roughly $118 neck-worn device powered by Alibaba's Qwen model that claims about 95% accuracy translating pet behavior, vocalizations and emotions into human speech.

AI-powered robot barber kiosks rolling out in Chinese cities scan the head in 3D and cut hair with millimeter precision for under a dollar per session.