GPT vs Grok: Who’s Actually Winning the AGI Race? (Honest Scoreboard)

Link to our newsletter: https://bitbiased.ai/
Every week there’s a new headline screaming that some model is "the smartest AI ever" and AGI is two years away. The charts contradict each other, the benchmarks are cherry-picked, and nobody measures the same thing twice.
So I skipped the hype reels and went through the actual technical reports, leaked parameter counts, benchmark numbers, and published roadmaps from OpenAI and xAI. The surprising part: the model winning the benchmark headlines isn’t necessarily the one closest to AGI.
This is GPT vs Grok across the three things that actually matter — what’s under the hood, what the benchmarks really say, and how close each one genuinely is to general intelligence. No hype, just the numbers we can verify.

Chapters:
00:00 The AI Headline Problem (Intro)
01:28 Model Architectures
03:34 Training Data & Compute
04:50 Benchmarks: GPT
06:08 Benchmarks: Grok
07:16 Key Takeaway on Benchmarks
08:00 AGI Indicators, Part 1: Generalization & Reasoning
09:28 AGI Indicators, Part 2: Tools, Learning & Safety
11:12 Summing Up the Indicators
11:41 Compute & Scaling
13:08 AGI Timeline Projections
14:02 The Probabilistic Forecast

If this breakdown saved you from another week of contradictory headlines, subscribe — skepticism is the whole point of this channel.

Who do you think cracks AGI first: the broad generalist or the brute-force specialist? Make your case in the comments.