Meta Launches Its New Flagship Image Model

Welcome back to The Shift. Let’s get straight to what matters in AI today…

Today we have:

🖼️ Meta Launches Its New Flagship Image Model

🗣️How to Dictate Into Any App on Your Mac With FluidVoice

🗣️ OpenAI Models Keep Expanding Their Reach

🔨Tools and Shifts you cannot miss

🖼️ Meta Launches Its New Flagship Image Model

Meta introduced Muse Image, the first image generation model from its Superintelligence Labs, bringing stronger image creation, editing, and agentic AI capabilities across Meta’s consumer products.

The Shift:

A Smarter Image Generator - Muse Image works alongside Muse Spark to search the web, use tools, refine its own outputs, and better understand complex creative requests before generating images.

Strong Benchmark Debut - The model ranked second on Arena’s text-to-image and image-editing leaderboards, placing behind only OpenAI’s GPT Image 2 in early public evaluations.

Rolling Out Across Meta - Muse Image is available free through Meta AI and is launching across Instagram and WhatsApp first, with Facebook, Messenger, and Meta Ads joining later.

Video Is Next - Meta also previewed Muse Video, its upcoming AI video model, which already ranks third on Arena’s video leaderboard behind Seedance 2.0 and Gemini Omni Flash.

Meta is expanding beyond chat assistants into foundation models for creative work.

By bringing advanced image generation directly into its apps, the company is strengthening its AI ecosystem while preparing to compete more aggressively in AI video creation.

Together with AutoSEO

SEO on your list now. Autopilot after setup

That SEO task that keeps getting pushed? AutoSEO handles it permanently. Articles, mentions, backlinks — running after setup without you touching it again. Over 2,500 small business owners went from "I'll get to it" to "it's already done."

Get my free plan

🗣️How to Dictate Into Any App on Your Mac With FluidVoice

3.7x faster than typing. Fully offline. Free forever. Here's how to set it up.

Step 1: Download FluidVoice Head to altic.dev/fluid and download the Mac app. Requires macOS 15 Sequoia or later. Apple Silicon and Intel both supported. Free, open source, GPLv3.

Step 2: Grant microphone and accessibility permissions FluidVoice needs mic access to capture audio and accessibility permissions to type into other apps. Both get prompted on first launch.

Step 3: Choose your speech model Pick from Nemotron Speech 3.5 for ultra-low latency, Parakeet TDT v3 for 25-language support, or Whisper for up to 99 languages. All run locally, nothing leaves your Mac.

Step 4: Enable Fluid-1 for smart post-processing Download the optional Fluid-1 model. It cleans up rough dictation, fixes casing, formats dates and numbers, and adapts tone per app, all on-device. No API key, no cloud.

Step 5: Set your tone per app Define custom prompts for each app. Slack gets casual, Mail gets formal, GitHub gets structured. FluidVoice detects which app is focused and switches automatically.

Step 6: Dictate anywhere Press your hotkey in any text field, email, Slack, Notion, terminal, Claude, VS Code. Speak naturally. Text lands in under 100ms.

50,000+ downloads. Free forever. You can download FluidVoice here.

🗣️ Open AI Models Keep Expanding Their Reach

Tencent and Kyutai announced major open-source AI releases this week, showcasing advances in efficient reasoning models and neural world simulation beyond traditional language generation.

The Shift:

Tencent Launches Hy3 - Tencent promoted Hy3 from preview to a full Apache 2.0 open-source release, claiming flagship-level performance while using far fewer active parameters and significantly less hardware.

Efficiency Over Size - Hy3 trails GLM-5.2 on coding benchmarks but performs strongly in web research and tool use, with a permissive license that avoids restrictions affecting earlier Chinese models.

A Game Without A Game Engine - Kyutai and General Intuition introduced MIRA, a neural world model that generates live 2v2 Rocket League entirely inside an AI model without relying on a traditional game engine.

Built From Observation - Trained on 10,000 hours of AI gameplay, MIRA learned the game without human data, rendering matches in real time while the teams open-sourced its code, dataset, and playable demo.

Open-source AI development is moving beyond larger language models.

The latest releases show researchers focusing on efficient reasoning, interactive world simulation, and broader accessibility, expanding what open AI systems can do.

Together with Maxio

The Ultimate Guide for Usage-Based Pricing for SaaS and AI

Implementing usage-based pricing successfully requires more than just a pricing strategy.

Download this guide for practical advice and best practices when considering usage-based pricing. 

👉 Get your guide

🔨AI Tools for the Shift

🏡 MeltFlex – Redesigns bedrooms, kitchens, bathrooms, offices, and other interiors with AI-generated layouts and furniture in seconds.

📚 BookRecall – Helps you remember and apply key ideas from books by resurfacing your highlights and insights when they are most useful.

🛒 Hostinger Ecommerce – Centralizes products, inventory, orders, and sales channels into one easy-to-manage ecommerce dashboard.

🧊 Create3D.io – Generates editable 3D models from text prompts or images with support for multiple export formats used in games, AR, ecommerce, and 3D printing.

📡 WiFi Analyser – Diagnoses Wi-Fi issues by analyzing PCAP files with hundreds of protocol checks and delivers root causes with vendor-specific fixes in seconds.

🚀Quick Shifts

🤝 Anthropic expanded Claude Cowork to mobile and web, enabling cloud-based AI work sessions across devices, background task execution, and mobile notifications, with rollout starting for Max subscribers. 

🚀 OpenAI’s GPT-5.6 model family has reportedly been cleared for public release, with flagship Sol, mid-tier Terra, and lightweight Luna models designed for high-performance, scalable, and everyday AI workloads.

🎨 Figma acqui-hired the team behind AI app-building platform Bud (formerly Orchids), strengthening its AI capabilities for designing, prototyping, and building applications with AI agents.

📱 Anthropic expanded Claude Cowork to mobile and web in beta, allowing users to seamlessly continue AI-assisted tasks across devices with cloud-synced sessions and background execution.

🏢 Microsoft is reportedly shifting Excel and Outlook AI prompts to its in-house MAI models, as AI chief Mustafa Suleyman pushes to reduce reliance on Anthropic and eventually eliminate the company's external model costs. 

🧩 Prompt of the Day

Explain Your Career Change With Confidence in Interviews

Most career changers focus on defending their past instead of connecting it to their future. A compelling story shows why you're changing, what you've done to prepare, and why you're ready now.

Turn your transition into a strength, not an obstacle.

Paste the prompt: Drop this into ChatGPT, then fill in your career details.

Prompt to paste

Help me prepare for a career transition from [Insert current career] to [Insert target career] within [Insert timeline]. My biggest challenge is [Insert biggest challenge]. Create a compelling interview narrative that explains why I'm changing careers, how my previous experience transfers to the new role, and what I've done to prepare for the transition. Include answers to common interview questions, strategies for addressing gaps in experience, examples of transferable skills, and recommendations for presenting my career change as a thoughtful, intentional decision rather than a risky leap.

🤳AI Shift of the Day

That’s all for today’s edition. See you tomorrow as we track down and get you all that matters in the daily AI Shift!

If you loved this edition, let us know how much:

Forward it to your pal to give them a daily dose of the shift so they can 👇