Are Grok 3 Benchmarks Fake?

Welcome to another edition of AI PlanetX.

Misleading Grok 3 benchmarks raise concerns; OpenAI rethinks reliance on Microsoft.

• Hottest AI News

• Top AI & SaaS Tools

• Top AI & Tech News

• Interesting Uses of AI: Prompt to Write Human-Like Content

• Top AI Video Tutorial

• AI Course of the Day: Intro to PyTorch & Neural Networks

Hottest AI News

xAI

xAI Under Fire Over Misleading Grok 3 Benchmarks

The AI community is abuzz as xAI faces claims of misleading benchmarks for Grok 3, especially on the AIME 2025 math exam, compared to OpenAI's models.

Details:

• xAI's graph showed Grok 3 beating OpenAI's o3-mini-high on AIME 2025 but omitted "cons@64" scores. Single-attempt scores reveal Grok 3 underperforms, contradicting xAI's "world's smartest AI" claim

• xAI co-founder Igor Babushkin defended their selective reporting, noting OpenAI has done the same. A neutral graph later showed Grok's cons@64 performance was solid despite the controversy

• The debate highlights AI benchmarking transparency issues. Costs and limitations behind scores remain unclear, showing benchmarks fail to fully convey model strengths and weaknesses

This incident underscores the need for standardized benchmarks and greater transparency in measuring and comparing AI capabilities.

There’s a reason 400,000 professionals read this daily.

Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.

Sign up now for free and work smarter, not harder.

OpenAI

OpenAI Plots Computing Switch From Microsoft to SoftBank

OpenAI is shaking up the tech world by diversifying its computing infrastructure beyond Microsoft. This strategic move could reshape the AI cloud computing landscape.

Details:

• OpenAI plans to shift 75% of its computing to the Stargate by 2030, reducing reliance on Microsoft Azure. Spending is projected to surge from 5B in 2024 to 20B by 2027

• Stargate Project, backed by SoftBank, is a $500B investment over four years to build US-based AI infrastructure. It could challenge cloud giants like Azure, AWS

• Microsoft updated its OpenAI partnership terms, adding a "right of first refusal" clause. This keeps Microsoft's priority intact while allowing OpenAI to collaborate with others

This shift is key as OpenAI transitions from a non-profit to for-profit, reaching a $157 billion valuation after raising $17.9 billion.

Top AI & SaaS Tools

• Linguix (Life-time Deal): Boost writing speed and quality with AI corrections, smart shortcuts, and context-based suggestions (Grammarly Alternative) [Price increases in 25 hours]

• Font Generator: World's first AI to generate custom fonts and download them in installable formats for Mac and Windows [F-R-E-E]

• Studio: ElevenLabs' AI text-to-audio editor lets you create audiobooks, voiceovers, and podcasts with pacing control, auto-voices, and GenFM [F-R-E-E]

• Chance iOS: Visual search engine—snap a photo of anything and explore its history, meaning, and connections. Android app launching shortly [F-R-E-E]

• HeyGen: Create professional, lifelike videos of yourself using a digital twin—no filming required. Ideal for those who are busy or camera-shy

Top AI & Tech News

• OpenAI bans accounts misusing ChatGPT for surveillance and influence campaigns

• Apple's Vision Pro to gain Apple Intelligence in April with visionOS 2.4 featuring AI voice assistants and predictive text

• Jensen Huang says investors got it wrong over DeepSeek stock selloff that wiped $600B from Nvidia

• Elon Musk's Grok 3 reportedly stated that both Musk and former President Donald Trump deserved the death penalty

• Qatar entered a 5-year agreement with Scale AI to implement AI tools and training to enhance its government services

Interesting Uses of AI

AI Art Spotlight

Model: Midjourney

Prompt:

Abstract painting of a horse, with a beige and blue color palette, brush strokes, and neutral colors. The background is beige, and the horse has a brown mane and a white face with black eyes. The style is minimalistic, with simple shapes and a watercolor-like appearance. --chaos 100 --ar 32:43 --quality 2 --sw 1000 --v 6.1

AI Prompt of the Day

Top AI Video Tutorial

5 AI Video Editing Apps You NEED to Try in 2025

Complimentary AI Course of the Day

Intro to PyTorch and Neural Networks

In this course, you will learn how to create, train, and test artificial neural networks in PyTorch, one of the most popular deep learning frameworks in Python.

THAT'S A WRAP