Google's Gemini 3 beats GPT-5 on reasoning

PLUS: Our exclusive interview with Demis Hassabis and Google's new free agent platform

Author

Gen AI Art
November 20, 2025 • Estimated Reading Time: 9 minutes

Together with

Google has just made a major play in the AI race, launching its next-generation Gemini 3 model. The new flagship AI is already claiming top performance on key reasoning benchmarks, directly challenging its top rivals.

With the simultaneous launch of a free platform for building AI agents, is this two-pronged strategy—a powerful model paired with accessible developer tools—the key to Google retaking the lead?

Today in AI:

  • Google's Gemini 3 tops GPT-5 on reasoning

  • Google's new free agent platform

  • Hassabis on AI-driven medical diagnostics

Introducing Voice AI Agents on WhatsApp

WhatsApp has always been where customers start conversations. Now, with Synthflow, those conversations can continue seamlessly over calls — answered directly by Voice AI Agents.

Enterprises can finally manage WhatsApp calls with the same automation, analytics, and security as phones.

The result: faster resolutions, 24/7 coverage, and a unified system for every customer call, whether it starts on telephony or WhatsApp.

What’s new? Google has launched Gemini 3, its next-generation flagship model, claiming top performance on key reasoning benchmarks and introducing a 'new era of intelligence'.

What matters?

  • Gemini 3, alongside its Deep Think architecture, surpasses GPT-5 on key reasoning benchmarks like Humanity’s Last Exam and also leads in scientific knowledge, math, and multimodal tasks.

  • Alongside the model, Google introduced Antigravity, a free, agent-first development platform that lets you orchestrate multi-agent workflows with browser control.

  • DeepMind CEO Demis Hassabis highlighted that Gemini 3’s powerful multimodal capabilities will serve as a foundation for specialized systems like AMIE, aiming to improve AI-driven medical diagnostics.

Why it matters?

Google has retaken a leading position in the AI race, directly challenging OpenAI's dominance with verifiable benchmark wins. The launch of Antigravity signals a major push toward an agent-based future, empowering developers to build more autonomous and capable applications.

GUIDE

What’s new? Alongside its new Gemini 3 model, Google launched Antigravity, a free platform designed to help developers build and manage complex AI agents.

What matters?

  • The platform is designed to simplify building AI agents, making it more accessible for developers to create applications that can perform tasks on their behalf.

  • It includes built-in browser control, which allows agents to navigate websites and interact with web-based applications to automate workflows.

  • Antigravity supports multi-agent orchestration, enabling developers to coordinate multiple specialized agents to tackle more complex, multi-step problems.

Why it matters?

Google is giving developers powerful tools to build the next wave of AI-powered assistants and automation. This move could accelerate the development of specialized agents that handle everything from complex research to daily administrative tasks.

PRESENTED BY SUPERHUMAN

Become the go-to AI expert in 30 days

AI keeps coming up at work, but you still don't get it?

That's exactly why 1M+ professionals working at Google, Meta, and OpenAI read Superhuman AI daily.

Here's what you get:

  • Daily AI news that matters for your career - Filtered from 1000s of sources so you know what affects your industry.

  • Step-by-step tutorials you can use immediately - Real prompts and workflows that solve actual business problems.

  • New AI tools tested and reviewed - We try everything to deliver tools that drive real results.

  • All in just 3 minutes a day

What’s new? In an exclusive interview, Google DeepMind CEO Demis Hassabis revealed how Gemini 3 will serve as a foundation for advanced applications, including AI-driven medical diagnostics.

What matters?

  • Hassabis envisions bringing the capabilities from specialized research systems like Project AMIE for medical diagnostics directly into the general Gemini model.

  • The model's powerful multimodal capabilities are a key reason for this direction, as many health-related questions involve interpreting complex, varied data types.

  • A long-term goal is to use the Gemini app on Android devices for democratizing healthcare by providing a basic level of medical knowledge in areas without sufficient primary care.

Why it matters?

The conversation highlights a significant move beyond benchmarks to applying AI to solve complex, real-world problems in high-stakes fields like medicine. Gemini 3's release signals a future where frontier models act as the core platform for a new generation of expert-level AI assistants.

Everything else in AI

Gemini 3 trails Anthropic's Claude Sonnet 4.5 on coding benchmarks, despite its record-breaking performance across most other reasoning and multimodal categories.

Google's research on its AMIE system showed the AI could achieve diagnostic accuracy on par with or even exceeding unspecialized physicians in text-based conversational assessments.

Google trained its new Gemini 3 models on its next-generation Tensor Processing Units (TPUs), highlighting the critical role of custom hardware in the performance of its new flagship model.

Google's new agent-building platform, Antigravity, is built on Firebase, allowing developers to leverage existing cloud infrastructure for creating and deploying complex AI workflows.

Essential AI Guides - Reading List:

Let us know!

Work with us

Reach 100k+ engaged Tech Professionals, Engineers, Managers and decision makers. Join brands like MorningBrew, HubSpot, Prezi, Nike, Ahref, Roku, 1440, Superhuman, and others in showcasing your product to our audience. Get in touch now →