Amazon is joining the reasoning race

Good morning, AI enthusiasts. The AI landscape is about to get much more competitive — with Amazon reportedly set to leap into the reasoning race with a robust hybrid model of its own.

With this latest reasoner coming alongside Alexa’s upcoming AI-powered upgrades, could this be the 12 months that finally establishes the retail giant as a real AI powerhouse?

In today’s AI rundown:

• Amazon’s hybrid reasoning AI model

• Cohere’s SOTA multilingual vision model

• Finding perfect prospects with AI

• OpenAI’s $50M NextGenAI consortium

• 4 latest AI tools & 4 job opportunities

LATEST DEVELOPMENTS

AMAZON

🧠 Amazon’s hybrid reasoning AI model

The Rundown: Amazon is reportedly developing a sophisticated reasoning AI model under its Nova brand—set for a June release—in what can be its most ambitious push yet to compete with OpenAI, Anthropic, and Google.

The small print:

• The corporate goals to create a “hybrid reasoning” system that delivers quick responses and methodical, multi-step problem-solving through a unified model.

• Cost-effectiveness is a central focus, with Amazon trying to undercut competitor pricing while still delivering top-tier performance.

• Amazon has reportedly set ambitious goals to rank among the many top five models, especially on benchmarks for software development and math skills.

• The project falls under Amazon’s AGI division led by Rohit Prasad—signaling a strategic shift despite the corporate’s massive $8B investment in Anthropic.

Why it matters: Amazon’s stake in Anthropic isn’t holding it back from developing its own rival models—aiming to compete in reasoning while undercutting each rivals and partners on price. With an AI-enhanced Alexa+ also within the pipeline, the retail giant is quickly positioning itself as a serious contender across multiple fronts within the AI race.

TOGETHER WITH UNSTRUCTURED

🎯 Improve RAG accuracy by 84%

The Rundown: Contextual chunking is the following level in document retrieval accuracy, and Unstructured goes to indicate you ways it beats Gemini 2.0’s long context windows each time—helping you say goodbye to high error rates and hello to specific RAG apps.

Join Unstructured’s live session to learn:

• How contextual chunking amplifies retrieval precision

• Comparisons with Gemini 2.0’s long context window approach

• Real-world strategies to cut back error rates and construct stronger RAG pipelines

Register here to hitch live or to get the recording.

COHERE FOR AI

🌐 Cohere’s SOTA multilingual vision model

The Rundown: Cohere’s non-profit research arm, Cohere For AI, unveiled Aya Vision, an open multimodal AI that brings vision-language capabilities to 23 languages representing over half the world’s population—setting latest performance benchmarks.

The small print:

• Aya Vision is available in two sizes, with the 8B version outperforming rivals 10x its size and 32B beating those greater than 2x its size, like Llama-3.2 90B Vision.

• The model can interpret and describe images, answer visual questions, and translate visual content across diverse languages—from Vietnamese to Arabic.

• It has been released under a CC non-commercial license and might be accessed on Kaggle, Hugging Face, or via WhatsApp.

• Cohere has also open-sourced the Aya Vision Benchmark, which evaluates VLMs on open-ended questions around real-world, multilingual scenarios.

Why it matters: We’ve previously written about AI models breaking down language barriers, and now, breakthroughs like Aya Vision are achieving the identical for visual content. Leveraging advanced AI won’t be limited to English-speaking audiences only, with users across the globe soon gaining access to a robust universal visual translator.

AI TRAINING

🔍 Finding perfect prospects with AI

The Rundown: On this tutorial, you’ll learn the best way to use Perplexity’s Deep Research feature to quickly generate focused B2B prospect lists and craft personalized outreach messages that convert higher.

Step-by-step:

1. Define your targeting criteria clearly (corresponding to industry, company size, location, and pain points).

2. Head over to Perplexity to generate prospects by entering a particular prompt requesting firms that match your criteria.

3. Ask it to prioritize and analyze your prospects by identifying probably the most promising matches and explaining why they’d profit out of your solution.

4. Generate personalized outreach messages for every priority prospect that reference their specific situation and funding.

Pro tip: You should utilize separate AI tools for research and content writing, e.g., take your prospect insights from step 3 and feed them into your favorite writing assistant for drafting polished outreach messages.

PRESENTED BY INNOVATING WITH AI

💼 Start your profession as an AI Consultant

The Rundown: Innovating with AI’s latest program, AI Consultancy Project, equips AI enthusiasts with all of the resources they should capitalize on the booming AI consulting market — which is ready to grow 8x to $54.7B by 2032.

This system offers:

• Tools and framework to search out clients and deliver top-notch services

• A 6-month roadmap to construct a 6-figure AI consulting business

• Student landing their first AI client in as little as 3 days

Click here to request early access to The AI Consultancy Project.

OPENAI

🎓 OpenAI’s $50M NextGenAI consortium

The Rundown: OpenAI announced NextGenAI, a brand new academic consortium backed by $50M in funding to support AI research and education across 15 leading institutions, including Harvard, MIT, and Oxford University.

The small print:

• The initiative provides research grants, compute resources, and API access to assist students, educators, and researchers advance high-impact AI applications.

• The partner institutions will tackle challenges from reducing rare disease diagnosis time to digitalizing historical texts and public domain materials.

• The consortium comes after OpenAI’s ChatGPT Edu launch last May, a reasonable version of GPT-4o created specifically for educational institutions.

• Notably, Perplexity can be moving in an identical direction, with eventual plans to make its Pro subscription free for college kids.

Why it matters: AI is on target to completely augment the standard scientific research and academic structures — and providing top institutions with each the resources and means to integrate the tech more widely into their workflows will help result in much more rapid advancements across the board.

QUICK HITS

🛠️ Trending AI Tools

• 🧮 Data Science Agent – Google’s latest tool for automating data evaluation

• 🗣️ Asyncflow v1.0 – Text-to-speech AI with over 450 voice options

• 🖥️ Browser Operator – Opera’s in-browser AI agent for completing tasks

• 🧠 Pieces Long-term Memory – AI agent to capture, resurface past workflows

💼 AI Job Opportunities

• 🛡️ Abridge – Governance, Risk and Compliance (GRC) Analyst

• ⚖️ OpenAI – Employment Counsel

• 🏗️ Scale AI – Enterprise Product Manager

• 📢 Metropolis – Marketing Coordinator

📰 Every part else in AI today

Google’s Pixel 10 will reportedly introduce “Pixel Sense”, an on-device assistant able to processing data from over 15 Google apps to finish different tasks.

Tencent’s Yuanbao AI app surpassed DeepSeek as the highest iPhone app downloaded in China this week, coming on the heels of its “fast-reasoning” Hunyuan Turbo release.

ASLP Labs introduced DiffRhythm, an open-weights model able to generating complete 4-min songs with vocals in only 10 seconds, using lyrics and magnificence prompts.

Amazon created a brand new dedicated agentic AI group inside AWS, with CEO Matt Garman calling it a “potential multi-billion business” that can help users automate tasks.

Cortical Labs launched CL1, the world’s first business “Synthetic Biological Intelligence” system that mixes living human brain cells with silicon hardware.

Cornell and Tel Aviv researchers created ProtoSnap, an AI that matches template characters to ancient cuneiform tablets, unlocking untranslated 3,000-year-old texts.

IPO-bound Coreweave announced the acquisition of AI developer platform Weights and Biases to integrate its capabilities into its cloud infra platform.

OpenAI CEO Sam Altman said GPT-4.5 will roll out to Plus users in phases “over just a few days” and suggested a credit-based system for accessing advanced features like Sora.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop this Friday at 3:30 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. You may learn the best way to use the most recent AI tools to take your coding to the following level, and the way we’re leveraging AI-powered coding at The Rundown.

RSVP here. Not a member? Join The Rundown University on a 14-day free trial.

🤝 Share The Rundown, get rewards

We’ll at all times keep this article 100% free. To support our work, consider sharing The Rundown with your mates, and we’ll send you more free goodies.

See you soon,

Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team