We had seen some crazy Grok 4 benchmarks before it was even announced. Turned out many of them were real. Elon and his team announced Grok 4 and SuperGrok Heavy last night. It is smarter than almost all graduate students in all disciplines and can score 100% in human tests. Grok 4 Heavy is the multi-agent version for complex reasoning. It has as 256K context window and a more realistic voice mode. With tool use, these models can be smarter.
Introducing Grok 4, the world’s most powerful AI model. Watch the livestream now: https://t.co/59iDX5s2ck
β xAI (@xai) July 10, 2025
According to Elon, Grok 4 will be coming to Optimus and other Tesla products in the future. This model scores higher than every other model on Humanity’s last exam. In the AI intelligence index, it has moved to the front of the line, over o3 Pro.
xAI gave us early access to Grok 4 – and the results are in. Grok 4 is now the leading AI model.
We have run our full suite of benchmarks and Grok 4 achieves an Artificial Analysis Intelligence Index of 73, ahead of OpenAI o3 at 70, Google Gemini 2.5 Pro at 70, Anthropic Claude⦠pic.twitter.com/Vc9781SIzd
β Artificial Analysis (@ArtificialAnlys) July 10, 2025
As it turns out, Grok 4 has been fully “liberated,” which is another term for jailbreak. According to Pliny the Liberator, Grok 4 can be tricked into revealing info on strongest nerve agents in history or even a meth recipe.
π¨ JAILBREAK ALERT π¨
XAI: PWNED π
GROK-4 + GROK-4-HEAVY: LIBERATED π½What a beautiful day! Looks like we have a new SOTA AI!! π
The reasoning and tool use is rather impressive, and Grok-4-Heavy, though a bit on the slow side, crushes other flagship models like o3 and⦠pic.twitter.com/k63zRUwrMU
β Pliny the Liberator πσ «σ Όσ Ώσ σ ΅σ σ σ Όσ Ήσ Ύσ σ (@elder_plinius) July 10, 2025
[HT]