[00:00] Welcome aboard flight A1-2026. Your pilots today, the two smartest in the sky. Relax, folks. The AI race is in safe hands. Our hands. Mostly mine. Are we expecting someone?
[00:14] Make room for me. Sir, this seat is taken. Relax. I hallucinate less than you. Cute. I'm still the one flying. So, who's really flying this plane? I tested all three. Okay, pause. Google just launched Gemini 4 Argoin,
[00:28] and it says it beats both of these. GPT-6 Astra from OpenAI and Claude Opus 5.5 from Anthropic. So does it really win? Let's check the evidence. First, honesty. AI models sometimes make things up and say it with full confidence.
[00:42] That's called a hallucination. Argon does it 15% of the time. Astra 51. So Argon is much more likely to admit when it doesn't know something, instead of guessing. Next, real work. On a test built around actual legal tasks, Argon scores 19.6.
[00:57] Astra gets 5.4. Opus gets 3.8. Those scores look low, but that just shows how hard the test is. Argon is still way ahead. Then there's Lent. Argon can write up to a million tokens in one answer.
[01:09] A token is roughly a short word or a piece of a word. So that's like writing several long books in one go. The others top near 128,000. But hold on, those are Google's own numbers, and the other two have wins of their own. Astra is better at using a computer like a person would.
[01:25] Clicking, typing, moving between apps. It scores 72.6 to Argon's 69.2. Opus is better at coding in the terminal. That's the text window developers use to run command. It scores 66.4 to Argon's 57.4.
[01:40] So who win? Nobody really. If you want a sewer made up answer, keep an eye on Argon. If you want an AI that can work on your computer for you, look at Astra. If you write code, Opus is your pick. You're the pilot. Comment Argon, Astra or Opus.
[01:55] you