k
keynews.ai
keynews.ai
FrontierThe story, in brief

Vercel collaborates with Google for Gemini 3 Pro Preview launch

17% correctness bump. Gemini 3 Pro Preview just landed in the top 2 on Vercel's Next.js evals—and it's already live in v0.

Reporting by Vercel Blog 1 min brief
Illustration of a transparent lens revealing connected networks across layers of paper.
Exploring the next frontier of AI research.AI illustration by KeyNews
The KeyNews take

Why it matters

Google's latest model shows material gains in coding and instruction-following, with early adoption signals from Vercel's developer tooling ecosystem. This matters for founders choosing inference partners and for understanding the competitive positioning of closed models on real-world coding tasks.

The key facts

6 to know

  1. Gemini 3 Pro Preview released today (Nov 18, 2025)

  2. 17% increase in correctness over predecessor on Vercel's Next.js evals

  3. Ranked in top 2 models on leaderboard

  4. Tested across Vercel's v0, Next.js, AI SDK, and Vercel Sandbox

  5. Improvements in instruction following, response consistency, coding focus, multimodal reasoning, and tool use

  6. Available via Vercel AI Gateway and v0.app

Go to the source

Vercel Blogvercel.com

Publisher excerpt: The Gemini 3 Pro Preview model, released today, is now available through the and on . Thanks to Google, Vercel has been testing Gemini 3 Pro Preview across v0, Next.js, AI SDK, and Vercel Sandbox over the past several weeks.Vercel AI Gatewayv0.app We've noticed the model has an increased focus on…
Read original report
Back to today's editionMore frontier news

The wider picture

View all
Illustration of a transparent lens revealing connected networks across layers of paper.
AI illustration by KeyNews
Frontier01

Alibaba's open-weight Qwen-Image-2.1 claims to beat closed models in image generation with just 7 billion parameters

A capable open-weight image model at 7B parameters challenges the closed-model dominance in generation and editing, expanding practitioner options for on-device and cost-efficient image workflows.

The Decoder
Illustration of a transparent lens revealing connected networks across layers of paper.
AI illustration by KeyNews
Frontier02

Tencent's Gander aims to keep talking while it works in the background

A novel architecture for multimodal agents that separates conversational continuity from task execution. Demonstrates a real capability tradeoff: smoother UX vs. task reliability. Relevant to how frontier labs are rethinking agent design.

The Decoder
Illustration of a transparent lens revealing connected networks across layers of paper.
AI illustration by KeyNews
Frontier03

Simulated students that make realistic mistakes help AI tutors learn faster

A novel approach to AI training using realistic synthetic feedback loops is accelerating tutor model development and reducing the cost of evaluation data. This represents a meaningful shift in how frontier labs can iterate on capability without massive labeled datasets.

The Decoder