Grok 3

Xai Unveils the Grok3 Language Model

4 min read4 days ago

--

In Short

Grok3 operates in three distinct modes:

Normal Mode: For everyday thinking and reasoning tasks.

Research Mode: Harnesses the power of the web, synthesising information from multiple sources to provide comprehensive answers.

Big Brain Mode: Tackles complex challenges with advanced problem-solving capabilities.

Research Mode: enhances your query by pulling in and integrating data from various web sources, ensuring you get the most informed and up-to-date responses.

Real-Time Reasoning: Similar to DeepSeek Chat, Grok3 displays its thought process in real-time, allowing you to follow along with its logic and reasoning as it happens.

Access Grok3through multiple platforms:
- Web Portal: Recommended for the best experience.
- Mobile App: Convenient for on-the-go interaction.
- X Platform: Seamlessly integrated for users of this social media environment.

For the most effective and feature-rich experience, xAI recommend using the web portal.

If you like this article & want to show some love ❤️

-Clap 50 times, each one helps more than you think! 👏

- Follow me on Medium and subscribe for free. 🫶

- Let’s connect on LinkedIn, and stay in touch on X!

Progress Made

Immense progress has been made by X.ai as shown in this graph, it would have been interesting to see the Grok3 plot here as well.

There are two Grok3 models, Grok3 and the Grok3 mini model. Below are benchmarking results. It feels like the models are reaching a point of marginal differences in performance.

So there are to factors at play here…the one is the single user market, where individuals subscribe to the models for primarily personal use and enterprise use (more about that later)

What is interesting is that four models are shown, Grok3, Grok3 mini, Grok3 Reasoning Beta and Grok3 mini Reasoning…

Benchmarking is really impressive, but with so many highly capable models, does it now come down to user experience being the differentiatior?

Subscriptions

Considering personal use, tiered subscriptions will play a big role…and that seems like something X.ai is trying to leverage, with extra features for subscribers to the Premium+ plan.

Or unlocking more features with a SuperGrok plan, carving out those marginal increments of users will to pay more.

Hence the focus comes down to user experience in terms of the UI, speed, modalities (X-platform, X.ai app and web) and if the product meets the expectation of the user.

UX

Grok has a DeepSearch function which is a research tool to synthesise a myriad of web sources into an answer. What is interesting, is how the process is surfaced and streamed in real time. You can see the chain of thought on the left, with the model’s internal reasoning on the right.

During the demo the question was put to Grok3 to plot out a mission from Earth to Mars and back…Grok generated the Python code for this task.

Enterprise Use

For enterprise use determining factors will be latency, how easily can be API be accessed in terms of preferred geographic and regional availability.

Speed, cost and API functionality.

Chief Evangelist @ Kore.ai | I’m passionate about exploring the intersection of AI and language. From Language Models, AI Agents to Agentic Applications, Development Frameworks & Data-Centric Productivity Tools, I share insights and ideas on how these technologies are shaping the future.

https://www.x.ai

--

--

Cobus Greyling
Cobus Greyling

Written by Cobus Greyling

I’m passionate about exploring the intersection of AI & language. www.cobusgreyling.com

No responses yet