Find your customers on Roku this Black Friday

As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. To that end, Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting options. After all, you know your customers, and we know our streaming audience.

Worried it’s too late to spin up new Black Friday creative? With Roku Ads Manager, you can easily import and augment existing creative assets from your social channels. We also have AI-assisted upscaling, so every ad is primed for CTV.

Once you’ve done this, then you can easily set up A/B tests to flight different creative variants and Black Friday offers. If you’re a Shopify brand, you can even run shoppable ads directly on-screen so viewers can purchase with just a click of their Roku remote.

Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

Use code GET5K now

Today:

• Claude Just Overtook Gemini 3—Here’s What That Means

• Black Forest Labs released FLUX.2: Stunning Image Gen with Creative Control

• ChatGPT Voice Now Includes Maps, Images, and Transcripts

• Trump Launches ‘Genesis Mission’ to Accelerate U.S. AI Innovation

• Claude AI Reduces Workload, Fuels Growth

Claude just beat Gemini 3... how?!

Anthropic’s Claude Opus 4.5 sets a new bar in AI performance, beating Google’s Gemini 3 Pro in coding (SWE-bench Verified 80.9 vs 76.2), agent tool use, and long-horizon tasks like VendingBench. It nears the threshold of automating junior AI research roles, excels as an orchestrator in multi-agent setups, and handles computer use with top scores. It even outperforms humans on Anthropic’s in-house engineering test. 

Though Gemini 3 Pro still leads in business simulations, Opus 4.5 is more efficient and empathetic in customer service tasks. Overall, it signals rapid, competitive progress across labs, proving AI development is accelerating on multiple fronts.

WATCH THE VIDEO ON YOUTUBE

Black Forest Labs released FLUX.2: Stunning Image Gen with Creative Control

FLUX.2 is a powerful, open-weight image generation and editing model built for real creative work. It supports high-res editing, consistent character and brand style, complex text rendering, and multi-image referencing—making it ideal for designers, developers, and companies seeking control, quality, and affordability.

KEY POINTS

• Creative Control & Quality: FLUX.2 generates 4MP photorealistic images with sharp textures, complex text, and layout accuracy, following detailed prompts and brand styles.

• Multi-Reference Editing: It supports up to 10 reference images to maintain character, product, and style consistency across outputs—ideal for storytelling, branding, and product shots.

• Open Innovation Ecosystem: Offers both open-weight and commercial models (FLUX.2 [dev], [pro], [flex], [klein]), supporting local deployment, community experimentation, and enterprise-level production workflows.

Why it matters

FLUX.2 brings powerful image creation tools to everyone—not just big companies. By sharing open models, it lets developers and artists build custom tools, save money, and create high-quality images that follow instructions well. It’s a big step for accessible, pro-grade visual AI.

ChatGPT Voice Now Includes Maps, Images, and Transcripts

ChatGPT Voice now works directly inside your ongoing chat. You can see a live transcript, view maps or images during the conversation, and no longer need to switch to a separate screen. A new "Separate mode" option is still available in Settings if preferred.

KEY POINTS

• Inline Voice Chat: Voice conversations now happen inside your existing chat window, with live transcription.

• Visual Aids Included: Maps, photos, and other visuals are shown alongside responses for better understanding.

• Optional Classic Mode: Users can return to the original full-screen voice interface by toggling “Separate mode” in Settings.

Why it matters

This makes ChatGPT Voice more natural and helpful. You can talk, see results, and follow along with visuals all in one place. It’s easier to learn, explore, and ask questions—especially when you’re looking for directions, recommendations, or explanations.

Trump Launches ‘Genesis Mission’ to Accelerate U.S. AI Innovation

President Trump signed an executive order launching the “Genesis Mission,” a government-wide initiative to accelerate artificial intelligence innovation. It aims to unify federal AI research efforts, integrate AI tools into scientific work, and drive faster breakthroughs across agencies, according to the White House.

KEY POINTS

• Genesis Mission Launched: New executive order promotes AI innovation across all U.S. government agencies.

• Centralized AI Strategy: Seeks to coordinate federal research efforts for more efficient development and deployment of AI tools.

• Focus on Science and Innovation: AI will be integrated into scientific missions to accelerate discoveries.

Why it matters

This marks a big push to make the U.S. government more AI-powered. By unifying research and applying AI to science, it could speed up innovation, strengthen national leadership in technology, and prepare agencies to handle future challenges with smarter tools.

🧠RESEARCH

General Agentic Memory Via Deep Research

A new memory system called GAM helps AI agents remember better by only keeping simple notes ahead of time and gathering deeper info when needed. This makes agents faster and more accurate. It uses two parts—a memorizer and a researcher—and improves with practice through reinforcement learning.

AutoEnv: Automated Environments for Measuring Cross-Environment Agent Learning

AutoEnv, a new benchmark framework, helps test how well AI agents can learn across different environments—not just in one familiar setting. It includes 36 diverse environments and shows that current learning methods don’t scale well. Adaptive strategies help, but gains drop off as complexity increases.

Unveiling Intrinsic Dimension of Texts: from Academic Abstract to Creative Story

This paper reveals that different writing styles have different levels of complexity for language models. Scientific texts are easier (low intrinsic dimension), while fiction and opinion pieces are harder (high intrinsic dimension). These differences aren’t linked to word predictability but to structure, tone, and emotion, offering new insights into how LLMs process language.

🛠️TOP TOOLS

Each listing includes a hands-on tutorial so you can get started right away, whether you’re a beginner or a pro.

AITubo – Quick Guide to Features, How It Works & Pricing - all‑in‑one creative suite for generating and editing images, videos, and music

AITWO – Quick Interior & Exterior AI Design - generative design platform for remodeling and visualizing interiors, exteriors, and architectural concepts

Aizzy AI – Privacy‑First AI Platform - all‑in‑one AI workspace designed for private, anonymous use

📲SOCIAL MEDIA

🗞️MORE NEWS

• Claude reduces task time by 80%, based on 100,000 real user chats. If broadly adopted, current AI could double U.S. productivity growth to 1.8% annually—mostly in tech, education, and professional jobs.

• Microsoft Edge now integrates Copilot AI to enhance holiday shopping. Users in the U.S. can access features like price comparisons, history, cashback alerts, and product insights—streamlining gift buying directly in the browser.

• Character.AI launched Stories, a choose-your-own-adventure feature for teens after banning open-ended chats. The new mode offers safe, structured storytelling with favorite characters, aiming to rebuild trust after serious safety concerns and lawsuits.

• Cloudflare now hosts FLUX.2, a powerful open-weight image model by Black Forest Labs. It generates photorealistic, customizable images with multi-reference support, JSON prompting, and superior character consistency—ideal for ads, product shots, and creative AI workflows.

• Nvidia shares dropped 4% after reports that Meta may use Google’s AI chips (TPUs) in its 2027 data centers. The move signals growing competition in AI hardware and demand for alternatives to Nvidia GPUs.

• OpenAI now lets ChatGPT Enterprise and Edu users choose where their data is stored, offering options in 10 global regions. This helps businesses meet local compliance rules, removing key barriers to wider AI deployment.

• Warner Music Group has partnered with AI music platform Suno, letting users create songs using voices and likenesses of artists who opt in. WMG dropped its lawsuit, signaling a broader industry shift toward AI licensing deals.

• MIT unveiled BoltzGen, a generative AI model that creates novel protein binders for hard-to-treat diseases. Unlike prior models, BoltzGen unifies design and prediction, shows strong performance on undruggable targets, and is fully open-source.