• Nextool AI
  • Posts
  • OpenAI's Models Are Making Chips Faster and More Efficient

OpenAI's Models Are Making Chips Faster and More Efficient

Plus: Snapchat Now Lets You Build AR Lenses Without Writing Code

In partnership with

OpenAI is using AI to design better chips, achieving up to 14% improvement in AMD's hardware designs. Meanwhile, Snapchat is making AR creation easier by letting developers build Lenses using Claude Code, Cursor, and Codex. And ResearchMonkey AI is tackling a different problem: trust in AI-generated research. Its platform connects summaries to original sources, helping students verify information instead of blindly trusting AI answers. Together, these developments show how AI is moving beyond simple assistance into engineering, creativity, and research.

In today’s post:

  • OpenAI's Models Are Now Designing Better AI Chips

  • Snapchat Just Made AR Creation a Lot Easier

  • AI Summaries Have a Trust Problem. This Tool Wants to Fix It.

SPONSORED BY

Frontier models perform better in Attio’s agent harness.

We gave GPT-6 Astra the same job twice: analyze 100 sales calls and draft the follow-ups, once in Ask Attio and once in ChatGPT.

Ask Attio finished in 4 minutes 49 seconds. ChatGPT took 26 minutes 11 seconds, about 5.4 times as long.

The most ambitious GTM work requires more than frontier intelligence. It requires architecture that can put that intelligence to work at scale.

What’s Trending Today

POLICY

OpenAI Is Using AI to Improve the Chips That Power AI

Image credits: OpenAI

OpenAI is pushing AI beyond writing code and generating answers. Its models are now helping engineers design faster, more efficient chips. OpenAI shared results from its chip design research. The improvements extend beyond its own custom hardware. Even AMD's highly optimized chips are seeing gains.

Here's what makes this development interesting.

  • AI is improving OpenAI's own chips: Models helped redesign components inside Jalapeño, OpenAI's custom inference chip. These changes reduced cell area by roughly 13% in its vector unit and 10% in its matrix unit.

  • AMD is seeing significant improvements: OpenAI reported up to 14% improvement in power, performance, and area metrics. These gains came from changes to AMD's flagship GPU designs.

  • Months of work are being compressed: OpenAI says the AMD collaboration achieved roughly a year's progress in under five months. Some AI-generated improvements have already entered manufacturing designs.

  • AI is working like an expert chip designer: Models rewrite hardware code to improve how circuits operate. Engineering tools then check functionality and measure performance trade-offs.

  • Newer models are getting better at hardware design: GPT-6.2 Astra led evaluations across two open-source RISC-V processors. It outperformed earlier models when optimizing existing designs.

  • Starting from scratch can work even better: Models also designed processor cores from written requirements. They achieved higher benchmark scores than when modifying existing designs, though simpler feature sets may explain some gains.

  • OpenAI is partnering with Synopsys: The companies are developing GPT-Synopsys for advanced chip design workflows. The goal is to help AI use professional engineering tools more effectively.

The interesting part isn't simply that AI can design chips. It's that AI is improving the hardware running AI itself. Better hardware could make future systems faster and more efficient. And better models could help design even better hardware. But benchmark improvements aren't the same as production-ready chips. Human verification and real-world testing still matter. The bigger question is how quickly this cycle can compound. Because AI might soon accelerate the development of its own infrastructure.

RESEARCH

Snapchat Now Lets AI Build AR Experiences for You

Image Credits: Snapchat

Snapchat is changing how augmented reality experiences are built. The company introduced Easy Lens CLI. It lets creators build AR Lenses using familiar AI coding assistants. Think Claude Code, Cursor, and Codex. The interesting part? You don't need to write code yourself.

Here's what makes this launch worth watching.

  • Build AR with simple prompts: Creators describe their ideas in plain English. AI assistants handle the technical work, from building to publishing.

  • Work with tools you already use: Developers can create Snapchat Lenses directly through Claude Code, Cursor, or Codex. There's less need to learn unfamiliar development tools.

  • Access powerful creative tools: Easy Lens CLI supports over 100 AR building blocks and 50 AI asset generators. That opens up more possibilities for creators.

  • Launch games in a single day: Snapchat says creators can turn game ideas into published Lenses within one day. AI handles testing, errors, and revisions.

  • Turn existing content into AR: Brands and agencies can transform campaign images, videos, and other assets into interactive experiences with less production work.

  • Earn money from your creations: Eligible developers can monetize Lenses through Snapchat's Lens+ Payouts program. Snapchat says it paid creators $1.8 million last month.

  • Build across operating systems: Easy Lens CLI works on Mac, Windows, and Linux. This makes the tool accessible to more developers.

The biggest shift isn't that AI can build AR experiences. It's that creators no longer need to master every technical step. Snapchat is making creation easier without removing creative control. But easier creation also means more competition. When anyone can build, good ideas become even more valuable. The real question is no longer who can build AR. It's who can create something people actually want to use.

BREAKTHROUGH

ResearchMonkey AI Wants You to Verify Every AI-Generated Answer

image credits: Research Monkey

ResearchMonkey AI is taking a different approach to academic research. Instead of just summarizing information, it links claims to their sources. That means students can check what AI actually gets right. And with AI hallucinations still a concern, that matters.

Here's what makes this platform worth watching.

  • Every summary comes with sources: ResearchMonkey AI attaches citations and exact quotations to its summaries. Users can trace claims back to specific pages, paragraphs, or timestamps.

  • Students are using AI, but trust remains shaky: A 2025 UK survey found 92% of undergraduates used AI tools. Yet 51% cited hallucinations as a concern.

  • Fake citations are a real problem: A 2025 study examined 176 GPT-4o-generated references. Nearly 20% were fabricated, highlighting the risks of unchecked research.

  • Research becomes easier to navigate: The platform summarizes PDFs, articles, presentations, and YouTube lectures. Each summary follows a consistent four-part structure.

  • Lectures and slides get detailed breakdowns: YouTube videos become searchable timelines. Presentations receive slide-by-slide explanations, helping students revisit important information.

  • Academic writing gets extra support: Students can generate essay outlines and organize research sources. Citation tools support APA, MLA, and BibTeX exports.

  • There's a free way to start: ResearchMonkey AI offers limited free access. Paid plans begin at $9.99 monthly, with a seven-day trial available.

AI can make research faster, but speed isn't everything. A convincing answer can still be completely wrong. ResearchMonkey AI is betting that verification matters more than convenience alone. Its citation features sound useful, though their reliability still needs testing. Because attaching a source doesn't automatically make a claim accurate. The better question is whether AI helps us think critically. Or simply makes it easier to stop checking.

Free Guides

My Free Guides to Download:

🚀 Founders & AI Builders, Listen up!

If you’ve built an AI tool, here’s an opportunity to gain serious visibility.

Nextool AI is a leading tools aggregator that offers:

  • 500k+ page views and a rapidly growing audience.

  • Exposure to developers, entrepreneurs, and tech enthusiasts actively searching for innovative tools.

  • A spot in a curated list of cutting-edge AI tools, trusted by the community.

  • Increased traffic, users, and brand recognition for your tool.

Take the next step to grow your tool’s reach and impact.

That's a wrap:

Please let us know how was this newsletter:

Login or Subscribe to participate in polls.

Reach 150,000+ READERS:

Expand your reach and boost your brand’s visibility!

Partner with Nextool AI to showcase your product or service to 140,000+ engaged subscribers, including entrepreneurs, tech enthusiasts, developers, and industry leaders.

Ready to make an impact? Visit our sponsorship website to explore sponsorship opportunities and learn more!