[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]  David Shapiro ⏩ [@DaveShapi](/creator/twitter/DaveShapi) on x 44.4K followers Created: 2025-07-24 19:03:49 UTC After realizing that SuperGrok here on X is NOT the same model as Grok X on grok-dot-com I've been able to revise up my opinion on the *real* Grok 4: Still not as smart as o3-pro. Not even close. It's definitely good at thinking about problems, quickly searching, and so on. It renders replies much faster, but for frontier research it does not add half as much to the conversation as o3-pro does. Here's some examples: talking about things like "cognitive hyper-abundance" and "intelligence too cheap to meter" as the result of the AI boom - Grok X excels at finding relevant results and explaining the basics. o3-pro still blows it out of the water, as it will add ideas that you did not explicitly ask about or think about. Grok X still feels sycophantic because it very much follows your lead like "yeah, yeah... you're probably right about that..." and doesn't offer much synthesis on its own. You still have to be VERY explicit about what you want it to think about - and it is very obedient once you fix the style. However, it doesn't give me any "autonomous agent" vibes. It's not like I'm talking to a true expert, it more like I'm talking to a really smart kid who's good at searching the internet and making arguments on a forum. Absolutely useful for some things, like fact-checking arguments and concepts, or learning about something well-established. Not useful for frontier research. Even though the average response time from Grok X seems to be XX to XX seconds, where o3-pro is still X to XX minutes, the information and insight density of o3-pro is *still* much more worthwhile. XXXXX engagements  **Related Topics** [faster](/topic/faster) [Post Link](https://x.com/DaveShapi/status/1948459118668513711)
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]
David Shapiro ⏩ @DaveShapi on x 44.4K followers
Created: 2025-07-24 19:03:49 UTC
After realizing that SuperGrok here on X is NOT the same model as Grok X on grok-dot-com I've been able to revise up my opinion on the real Grok 4:
Still not as smart as o3-pro. Not even close.
It's definitely good at thinking about problems, quickly searching, and so on. It renders replies much faster, but for frontier research it does not add half as much to the conversation as o3-pro does.
Here's some examples: talking about things like "cognitive hyper-abundance" and "intelligence too cheap to meter" as the result of the AI boom - Grok X excels at finding relevant results and explaining the basics.
o3-pro still blows it out of the water, as it will add ideas that you did not explicitly ask about or think about. Grok X still feels sycophantic because it very much follows your lead like "yeah, yeah... you're probably right about that..." and doesn't offer much synthesis on its own. You still have to be VERY explicit about what you want it to think about - and it is very obedient once you fix the style.
However, it doesn't give me any "autonomous agent" vibes. It's not like I'm talking to a true expert, it more like I'm talking to a really smart kid who's good at searching the internet and making arguments on a forum. Absolutely useful for some things, like fact-checking arguments and concepts, or learning about something well-established.
Not useful for frontier research.
Even though the average response time from Grok X seems to be XX to XX seconds, where o3-pro is still X to XX minutes, the information and insight density of o3-pro is still much more worthwhile.
XXXXX engagements
Related Topics faster