Kind: captions
Language: en
Everyone's asking me which AI company
will hit AGI first, and most people are
betting on OpenAI because they're the
household name. But here's what no one's
talking about. Elon Musk just told his
XAI employees there's a 10% chance their
next model achieves AGI and Sam Alman
quietly stated that OpenAI already knows
how to build it. I've been tracking
every leak, road map, and insider report
from both companies, and what I found
shocked me. One of them is moving three
times faster than anyone realizes.
Welcome back to bitbiased.ai,
where we do the research so you don't
have to. Join our community of AI
enthusiasts with our free weekly
newsletter. Click the link in the
description below to subscribe. You will
get the key AI news, tools, and learning
resources to stay ahead. So, in this
video, I'm breaking down the actual road
maps for both companies. OpenAI
currently has GPT 5.2, 2, but they're
working on GPT 5.3 and GPT6 behind the
scenes. XAI just released Grock 4.1, but
Grock 4.2 and the massive Gro 5 are
coming within weeks.
I'll show you the technical strategies
each is using, what separates them, and
give you my honest prediction on who
crosses the AGI finish line first.
Let's start with where OpenAI is right
now and what they're building next.
OpenAI strategy, the methodical march to
AGI. Sam Alman dropped a bombshell
statement recently that most people
glossed over. He said, "Open AI is now
confident we know how to build AGI as we
have traditionally understood it." Think
about that for a second. The CEO of the
leading AI company just said they've
cracked the code. Not that they're still
figuring it out, not that they're making
progress, but that they know how to do
it.
But here's where it gets interesting.
Open AI isn't rushing. They're taking
what Altman calls an iterative and
safetyconscious approach. Every model
they release is a stepping stone,
allowing society to adapt while they
refine the technology.
It's like they're building AGI the way
you'd assemble a rocket ship. Carefully,
piece by piece, testing every component
before launch. Their whole philosophy
boils down to this. scale up
capabilities, but never lose sight of
alignment and safety.
They've watched enough sci-fi movies to
know what happens when you build
something too powerful without making
sure it plays nice. That's why they're
pouring massive resources into alignment
research, making sure that when AGI
arrives, it actually does what humans
want it to do. Right now, OpenAI's
flagship is GPT 5.2, which is publicly
available. This GPT5 series introduced
something fascinating. A unified system
that can dynamically switch between
modes. Think of it like having two
brains. One that's lightning fast for
simple tasks and another that kicks in
for deep complex reasoning. GPT5 even
has a router that decides in real time
which mode to use. It's not just about
making models bigger anymore. It's about
making them smarter about using their
intelligence. and the investment backing
this absolutely staggering. We're
talking hundreds of billions of dollars
going into data centers and computing
infrastructure.
Altman himself said they've never hit a
ceiling where more computing power
didn't translate to better performance.
Double the compute, roughly double the
capability.
That's why they're building
supercomputers that would make NASA
jealous.
GPT5.3,
the rumored upgrade.
everyone's whispering about. Now, here's
where things get spicy.
OpenAI hasn't officially announced GPT
5.3 yet, but the AI community is
absolutely buzzing.
According to multiple leaks and insider
reports, there's a version code named
garlic in development right now. Yes,
they apparently named it after a
vegetable, and it's expected to be a
strategic upgrade over the current GPT
512. What kind of improvements are we
talking about? First, a massively
expanded context window.
Imagine being able to feed an entire
novel into chat GPT and have it remember
every detail, every character arc, every
subplot without losing the thread.
That's the direction Open AI is heading.
The rumors also mention enhanced
long-term memory across sessions, which
could be game-changing for anyone using
AI as a work assistant.
Instead of constantly reexplaining your
project context every single
conversation, the AI would just remember
what you told it last week. There's also
talk of developer tools that sound
straight out of science fiction. Things
like secure tunnels for the model
context protocol, which would let
external applications feed data into the
model safely. In practical terms, GPT
5.3 could become the AI that actually
sticks with you through extended
projects using tools and external data
sources without forgetting what it
learned 5 minutes ago. Now, I need to
emphasize OpenAI is still officially
standing behind GPT 5.2 as their latest
flagship.
Everything about GPT 5.3 is unconfirmed
speculation from industry insiders and
leakers. But here's why these rumors
make strategic sense.
Look at the competition. Google's
Gemini, Anthropics Claude, Meta's Llama,
they're all pushing hard on long context
windows, tool use, coding capabilities,
and agent-like behavior.
Open AAI can't afford to stand still.
Even if GPT 5.3 never gets that exact
name, you can bet they're working on
iterative improvements behind closed
doors to stay ahead.
Think of GPT 5.3 as open AI shoring up
their foundations before the big leap.
They're making sure their current
flagship is bulletproof, reliable, and
ready to handle realorld work at scale.
GPT6, the model that could change
everything. And then there's GPT6. This
is where things get really wild. Sam
Alman has been koi about specifics, but
he dropped hints in late 2025 that are
hard to ignore. He said they expect
significant gains from GPT 5.2 in the
first quarter of 2026.
We're talking about something so
advanced it might warrant jumping to a
whole new generation number. In other
words, GPT6 could be arriving right now.
So, what might GPT6 actually look like?
Based on OpenAI's trajectory and
Altman's philosophical musings, we can
make some educated guesses. The goal is
clear. Get as close to AGI as possible.
an AI that doesn't just excel at
specific tasks, but can genuinely learn
and think its way through anything you
throw at it. But here's what's
fascinating about Alman's vision.
He's acknowledged that AGI is an
incredibly vague term. Nobody really
agrees on what it means. So he offered a
thought experiment.
True super intelligence might be an AI
that could do a better job as president
of the United States than any human,
even if that human had AI assistance.
It's bold, provocative, and a little
terrifying,
but it shows the level of capability
they're ultimately aiming for.
GPT6 will almost certainly require a
completely different scale of computing
power.
We're talking about training runs that
might cost billions of dollars. The
model could incorporate explicit memory
modules, advanced reasoning tools, or
even be a collection of specialized
expert systems working in concert.
OpenAI has already experimented with
multi-system architectures in GPT5. So
GPT6 could take that to the extreme. And
wait until you hear this. Alman wrote in
early 2025 that we may see the first AI
agents join the workforce and materially
change the output of companies sometime
this year.
He's not talking about chat bots. He's
talking about autonomous AI systems that
can read, write, code, plan, and execute
complex multi-step tasks across
different domains.
GPT6 might be the engine that powers
those agents.
But OpenAI isn't just racing blindly
toward AGI. They're acutely aware of the
risks. They've talked about needing
regulation for super powerful systems,
similar to how we handle nuclear
technology.
Altman has said open AI cannot be a
normal company given what's at stake.
They're exploring governance mechanisms
to ensure AGI remains beneficial, not
destructive. The bottom line, GPT6 looms
as the potential gamecher. It could be
the model where OpenAI finally achieves
something they genuinely call AGI. And
with Sam Alman's combination of massive
computational resources and a measured
gradual deployment strategy, they're
positioning themselves to be first
across the finish line, but only if they
can do it safely. Elon Musk's XAI, the
bold, fast-moving challenger.
Now, let's switch gears and talk about
the wild card in this race, Elon Musk
and XAI. If OpenAI's approach is steady
and measured, XAI is the exact opposite.
It's bold, aggressive, and moving at
absolutely breakneck speed. Elon Musk
co-founded OpenAI back in 2015, but left
a few years later. Now he's back in the
AI arena with his own company, and he's
not here to play second fiddle. Musk
framed XAI's mission in the most Elon
way possible, to understand the
universe.
That's not marketing speak. He literally
wants to build an AI that can help
answer the deepest questions about
reality.
His mantra for XAI is what the hell is
really going on? It's philosophical,
ambitious, and quintessentially Musk.
And here's what makes XAI terrifying for
Open AI. They're moving fast.
XAI currently has Gro 4.1 publicly
available, but Grock 4.2 2 is dropping
in just weeks and the massive Gro 5 is
already in training.
While OpenAI carefully plans each
release, Musk is iterating at a pace
that seems almost reckless.
The XAI philosophy, truth over political
correctness. Here's where XAI diverges
from everyone else. Musk has openly
criticized models like ChatGpt for being
too filtered, too politically correct.
He wants Grock, that's XAI's AI model,
to tell it like it is, even if the truth
is uncomfortable or doesn't fit
corporate friendly narratives. The idea
is that to truly understand the
universe, an AI needs freedom to explore
truth without restrictive guard rails.
This has resulted in Grock having what
Musk calls a rebellious streak and even
a sense of humor.
When early testers got access, they
found an AI that was sarcastic, witty,
and willing to engage with topics that
other AI companies shy away from.
It reflects Musk's own online persona in
a lot of ways. But XAI strategy goes way
beyond personality. They're leveraging
Musk's entire empire of companies.
Unlike OpenAI, which started as a pure
research lab, XAI is integrated into
Twitter, now X, Tesla, and potentially
SpaceX from day one. They're training
Grock on the massive fire hose of public
tweets, giving it real-time knowledge of
current events and pop culture. And
they're deploying Grock directly into
Tesla vehicles. You can already use the
AI assistant in your car for navigation,
questions, or entertainment via voice
commands. This distribution strategy is
genius. Tesla has millions of customers
worldwide. By putting Grock in every
Tesla, XAI instantly reaches a massive
audience and gets invaluable real world
feedback to improve the model. And then
there's the money. Musk has secured $20
to $30 billion per year in funding for
XAI's computing needs. NVIDIA and AMD
have both invested.
XAI's valuation briefly hit $230
billion, making it more valuable than
OpenAI for a moment. The message is
clear. Musk is going allin on an
infrastructure arms race. Biggest
computers, largest models, fastest
iteration. He's even speculated about
building data centers in space staffed
by Tesla's humanoid robots to get an
edge in computing capacity. It sounds
insane, but it's classic Elon.
Grock 4.1, where XAI stands today. Let's
talk about where XAI is right now with
Grock 4.1.
First, the naming scheme. You might
notice these version numbers have a
pattern. 4.1 eventually leads to 4.2,
which hints at 42, the famous answer to
life, the universe, and everything from
The Hitchhiker's Guide to the Galaxy.
Musk gave his AI model version numbers
that build toward a geeky joke about it
answering everything. And Grock itself,
it's a sci-fi term meaning to understand
something deeply and intuitively. The
name alone sets an incredibly high bar.
Grock 4.1 is built on the foundation of
Gro 4, which launched mid 2025 and
represented a massive leap forward.
We're talking about a 100fold increase
in training compute compared to the
previous generation. XAI scaled up to
roughly 1 million GPUs for training.
That's an unfathomable amount of
processing power. But it's not just
about size. The Gro 4 series introduced
something called Gro 4 heavy, a special
mode where the model runs multiple
reasoning threads in parallel,
considering different hypotheses
simultaneously.
And it worked. Gro 4 Heavy became the
first model to score above 50% on
humanity's last exam, a benchmark so
difficult that most AI models fail
spectacularly.
Scoring even half the points was
unprecedented. It demonstrated reasoning
abilities approaching expert human
problem solving in some domains.
Grock 4.1 which is currently available
has multimodal capabilities.
There's a feature called Grock imagine
that handles image generation and
understanding. The context window is
massive 256,000 tokens. That's eight
times larger than GPT4's maximum. With a
context window that huge, Grock can
ingest entire books, technical manuals,
or research papers in one go and reason
across all of it simultaneously. But
here's what really sets Grock apart from
open AI. Real-time knowledge and live
tool use.
Grock integrates a live search API that
pulls up totheminute information from
the web exit and news sources.
This means Grock can answer questions
about events that happened 5 minutes
ago. And XAI has tested this in
highstakes scenarios. They ran Grock in
a simulated stock trading competition
called Alpha Arena where it analyzed
real-time market data and made trading
decisions.
Grock achieved an average 12% return and
ranked number one among all AI
competitors.
That's the difference in philosophy.
While OpenAI focuses on benchmarks and
safety, XAI is proving Grock can excel
in live, dynamic, realworld applications
where millions of dollars could be on
the line. Grock 4.2, the imminent
upgrade. But XAI isn't stopping with
4.1. Gro 4.2 is coming very soon. Elon
Musk stated in December 2025 that it
would launch in about 3 weeks, which
puts us right at the launch window now
in late January 2026.
What's expected in 4.2? According to
insider leaks, this will be a polished
refinement with enhanced multimodal
features and even better real-time
capabilities.
The improvements seem focused on both
raw capability and usability.
XAI went from Gro 4.0 0 to 4.1 in just a
few months and 4.2 is arriving weeks
later. That's an unprecedented rate of
major model releases.
Musk is clearly pushing his team to
catch up with and potentially leapfrog
open AI through sheer velocity.
Gro 5, the AGI moonshot.
And now we get to the big one. Gro 5.
Elon Musk has been incredibly bullish
about what Grock 5 represents. He's told
XAI employees and even announced on
social media that Gro 5 is aimed
squarely at achieving AGI. In one all
hands meeting, Musk put a specific
number on it. There's roughly a 10%
chance that Gro 5 will achieve the
world's first AGI. Now 10% might not
sound high, but think about what he's
saying. He's claiming even a 1 in 10
shot at actual AGI in the next
iteration. That's either tremendous
ambition or incredible showmanship.
probably both. And Musk expects Grock 5
very soon in December 2025. He said it
would be ready within a few months.
Based on the road map, we're talking
about January 2026.
It could drop any day now. What makes
Gro 5 potentially revolutionary?
The details point to sheer scale and
entirely new capabilities.
Gro 5 is reportedly built on a model
with around 6 trillion parameters.
Let me put that in perspective.
6 trillion is orders of magnitude larger
than the 100 billion parameter models
that were state-of-the-art just 2 years
ago. If accurate, Gro 5 would be the
largest AI model ever publicly
announced. Musk's theory is that this
scale combined with advanced training
methods and possibly mixtures of expert
subsystems could be the key to unlocking
AGI.
It's not just about making it bigger for
the sake of being bigger. It's about
reaching a threshold where the model can
do something qualitatively different.
Learn, reason, and innovate in ways
current AI cannot.
And Musk has hinted at capabilities that
sound like science fiction. He suggested
Gro 5 might be capable of making new
scientific discoveries, even discovering
new physics. That's speculation
obviously, but it signals what XAI is
shooting for. An AI that doesn't just
consume human knowledge, but generates
entirely new insights from first
principles. If Grock 5 can do even a
fraction of that, it would absolutely
earn the AGI label. There's also the
competitive angle. Musk is positioning
XAI as a direct rival to OpenAI and
Google. He's been candid that he started
XAI because he was worried about OpenAI
and Google racing ahead without enough
oversight. By creating XAI, Musk wants
to build what he calls a good AGI,
one that's aligned in the way he
believes is best for humanity.
This possibly means an AGI that's more
transparent, whose code might be more
open, and that prioritizes values like
truth seeeking and freedom of speech.
Musk has also been vocal about AI risks
and the need for regulation.
He's met with government officials and
world leaders to discuss AI. So, he's
not charging ahead recklessly.
But in classic Musk fashion, he's
balancing caution with an intense drive
to innovate faster than anyone else.
Beyond Grock 5, XAI has fascinating
projects in the pipeline. They're
working on deep Xplatform integration,
weaving Grock into the entire Twitter UX
ecosystem.
Imagine scrolling your feed and having
an AGI agent summarize discussions, fact
check claims in real time, or curate
personalized content.
Musk is creating a world where Grock is
everywhere, in your car, on your social
media, potentially in robots. Tesla's
Optimus humanoid robot could use Grock
as its brain.
Musk has even alluded to Optimus robots
powered by Grock potentially working on
Mars outposts someday.
This synergy means if XAI achieves AGI
with Gro 5, it would immediately have a
massive platform to deploy in the real
world. So that's XAI's vision. Gro 5 as
a moonshot for AGI built on trillions of
parameters and Musk's confidence that
brute force compute plus ingenuity
equals generally intelligent systems.
They're moving at breakneck speed. and
Musk's bold predictions, AGI by 2026,
surpassing human intelligence shortly
after, reflect both the confidence and
the astronomical stakes of this venture.
The verdict, who's actually closer to
AGI?
So, here we are at the million-doll
question. Which path is closer to
achieving AGI?
It's honestly like watching two Formula
1 drivers take completely different
strategies in the same race. Both are
capable of winning, but their approaches
couldn't be more different. Let me break
down where each actually stands right
now. Open AAI has GPT 5.2 in production
with rumored GPT 5.3 improvements in
development and GPT6 as their big AGI
bet, potentially arriving in early 2026.
XAI has Gro 4.1 publicly available with
Gro 4.2 2 launching within weeks and Gro
5, their AGI moonshot with 6 trillion
parameters expected to drop any day now.
The pace difference is staggering. While
OpenAI carefully plans each release with
extensive safety testing, XAI is
shipping major updates every few weeks.
Musk told his employees there's a 10%
chance Gro 5 achieves AGI.
Meanwhile, Altman says OpenAI already
knows how to build AGI. They just need
to execute carefully. OpenAI's approach
is characterized by careful balance.
Innovation meets caution. They have a
proven track record with GPT4 and GPT5.
Their culture is systematic, methodical
research. Sam Alman and his team are
philosophical about it. Push the
frontier, yes, but emphasize alignment,
safety, and gradual releases.
Their advantage, they're currently the
market leader. Their models are
everywhere, used by millions, trusted by
developers and businesses worldwide.
If AGI requires not just massive scale,
but also careful fine-tuning of
behavior, safety mechanisms, and
alignment with human values, OpenAI's
years of experience are invaluable.
When Altman says OpenAI knows how to
build AGI, he's signaling that all the
pieces are in place. Now it's about
engineering at scale and solving the
final tough problems methodically. XAI's
approach is pure ambition and
acceleration. Elon Musk is trying to
leaprog everyone through overwhelming
force. Build the biggest computers,
train the largest model, integrate it
everywhere to gather feedback at
lightning speed. Their newcomer status
hasn't stopped them from iterating
insanely fast and hitting impressive
benchmarks.
If AGI can be achieved by throwing 10x
or 100x more compute and data at the
problem than anyone has before, XAI
might get there first.
Here's what's fascinating. X AAI is
unconstrained in ways Open AI isn't.
They're willing to try unconventional
approaches, train on unfiltered data,
embrace real-time learning, and move
fast without as much baggage from years
of safety committees.
Musk's prediction that XAI could achieve
AGI in 2026 shows his confidence in this
high-speed, high-risk approach. So,
who's closer? Here's my honest take.
Open AAI is closer in terms of
refinement, reliability, and proven
capabilities.
They're building AGI the way you'd build
a space station. Carefully with
redundancies, making sure every system
works before you add the next module.
XAI is closer in terms of raw speed,
willingness to take risks, and sheer
compute power being thrown at the
problem.
They're building AGI like Musk builds
rockets. Iterate fast, test
aggressively, and accept that some
things might explode along the way. Open
AAI's GPT6 could be the first model they
consider AGI complete.
XAI's Gro 5 is aiming for that title,
possibly even sooner, potentially within
weeks.
Musk has positioned XAI as the underdog
that could shock the world. If Grock 5
lives up to even half of its billing, if
it matches or exceeds GPT 5.2's
performance across a broad array of
tasks while demonstrating real
autonomous learning, it could put XAI
neck andneck with OpenAI overnight. But
here's the thing, AGI isn't just a
benchmark score. It's a qualitative leap
to an AI that's truly general, that can
learn like we learn, reason like we
reason, and maybe even surpass us.
Both companies might be exploring
complimentary routes to the same
destination.
Open AAI might achieve an aligned safe
general intelligence by carefully
assembling every puzzle piece. XAI might
achieve AGI by building a massive engine
and stress testing it in real world
scenarios until something extraordinary
emerges.
And let's be real, there's the
possibility neither gets there first.
Deep Mind, Anthropic, or some stealth
startup we're not watching could steal
the crown. But right now, Open AAI with
their refinement and XAI with their
velocity are absolutely the front
runners. If I had to place a bet today,
I'd say OpenAI's depth of research and
proven track record gives them a slight
edge in robustly realizing AGI. An AGI
that's trustworthy, safe, and society
integrated. But XAI's audacity and raw
speed gives them a legitimate shot at
crossing some AGI milestone first, even
if it's rougher around the edges.
Altman's approach might yield an AGI we
can trust. Musk's approach might yield
an AGI that's frighteningly powerful and
capable.
Which of those counts as the real AGI
first? We're about to find out. What's
absolutely clear is that we're living in
one of the most extraordinary moments in
human history.
AGI was pure science fiction just a few
years ago. Now, we have the leaders of
top AI companies openly saying it could
arrive within weeks or months. Sam Alman
reflects on how surreal it is that we
went from the quiet launch of chat GPT
to the brink of AGI in just a few years.
Elon Musk is betting that AGI is the
most transformative technology in human
history, which means you go allin to
both create it and control its
trajectory.
Both Open AI and XAI are climbing the
same mountain from different sides.
OpenAI's trail is wellmapped, safety
roped, and tested.
XAI's trail is steep, dangerous, and
blazingly fast. They might converge near
the summit, or one might reach it first
while the other is still securing their
equipment.
For those of us watching this unfold in
real time, it's thrilling and honestly a
little terrifying.
We're not talking about incremental
improvements anymore. We're talking
about models that could fundamentally
change what work means, what thinking
means, and what intelligence means.
Whether AGI arrives via GPT6, Gro 5, or
something else entirely, the world will
be fundamentally different on the other
side of that threshold.
The stakes couldn't be higher, and the
timeline has never been shorter.
Thanks for sticking with me through this
deep dive. Drop a comment below with
your prediction.
Open AAI or XAI, who crosses the AGI
finish line first? And if you want to
stay on top of this race as it unfolds,
subscribe because we're tracking every
development. These companies are moving
fast and things could change literally
overnight.