did the title say Claude 3 this whole time? 🤦
Wes Roth
286K subscribersThe latest AI News. Learn about LLMs, Gen AI and get ready for the rollout of AGI. Wes Roth covers the latest happenings in the ...
did the title say Claude 3 this whole time? 🤦
46 Comments
TheAIGRID
99.5k views • 1 day ago
Wes Roth
91.1k views • 1 day ago
Wes Roth
11.6k views • 1 day ago
AI For Humans
2.1k views • 2 days ago
Wes Roth
79.9k views • 2 days ago
TheAIGRID
21.2k views • 2 days ago
Wes Roth
10.0k views • 2 days ago
AI For Humans
25.0k views • 3 days ago
TheAIGRID
13.6k views • 3 days ago
Wes Roth
80.3k views • 3 days ago
Shelf will be hidden for 30 daysUndo
Wes Roth
13.8k views • 3 days ago
Wes Roth
24.6k views • 4 days ago
TheAIGRID
8.1k views • 4 days ago
AI For Humans
8.9k views • 5 days ago
Wes Roth
29.3k views • 5 days ago
TheAIGRID
17.6k views • 5 days ago
TheAIGRID
49.3k views • 1 week ago
AI Jason
31.5k views • 1 week ago
Wes Roth
19.4k views • 1 week ago
Wes Roth
15.0k views • 1 week ago
Wes Roth
201.6k views • 1 week ago
TheAIGRID
32.6k views • 1 week ago
Wes Roth
148.3k views • 1 week ago
TheAIGRID
3.3k views • 1 week ago
TheAIGRID
23.5k views • 1 week ago
Wes Roth
45.1k views • 1 week ago
AI Explained
90.5k views • 1 week ago
Wes Roth
17.4k views • 1 week ago
AI For Humans
18.8k views • 1 week ago
Wes Roth
4.9k views • 1 week ago
AI Jason
16.1k views • 1 week ago
Wes Roth
51.5k views • 1 week ago
TheAIGRID
439.3k views • 1 week ago
AI Explained
95.2k views • 1 week ago
Wes Roth
103.7k views • 1 week ago
AI For Humans
25.6k views • 1 week ago
Wes Roth
53.8k views • 1 week ago
AI Jason
7.6k views • 1 week ago
TheAIGRID
31.1k views • 1 week ago
AI For Humans
1.8k views • 1 week ago
AI Explained
74.6k views • 1 week ago
Wes Roth
62.3k views • 1 week ago
TheAIGRID
37.3k views • 2 weeks ago
Wes Roth
9.5k views • 2 weeks ago
Wes Roth
3.0k views • 2 weeks ago
Wes Roth
5.9k views • 2 weeks ago
Wes Roth
13.9k views • 2 weeks ago
TheAIGRID
39.8k views • 2 weeks ago
AI For Humans
13.8k views • 2 weeks ago
Wes Roth
99.4k views • 2 weeks ago
Wes Roth
58.1k views • 2 weeks ago
AI For Humans
1.5k views • 2 weeks ago
TheAIGRID
9.6k views • 2 weeks ago
TheAIGRID
35.8k views • 3 weeks ago
AI For Humans
21.2k views • 3 weeks ago
AI For Humans
1.3k views • 3 weeks ago
TheAIGRID
29.3k views • 3 weeks ago
AI Jason
24.4k views • 3 weeks ago
AI For Humans
1.5k views • 3 weeks ago
AI For Humans
1.7k views • 3 weeks ago
TheAIGRID
11.6k views • 4 weeks ago
TheAIGRID
24.4k views • 1 month ago
AI For Humans
12.7k views • 1 month ago
TheAIGRID
13.6k views • 1 month ago
TheAIGRID
22.0k views • 1 month ago
TheAIGRID
7.6k views • 1 month ago
TheAIGRID
39.1k views • 1 month ago
AI Explained
98.3k views • 1 month ago
TheAIGRID
27.7k views • 1 month ago
AI Explained
60.0k views • 1 month ago
AI For Humans
17.2k views • 1 month ago
TheAIGRID
5.9k views • 1 month ago
TheAIGRID
50.8k views • 1 month ago
AI Jason
45.3k views • 1 month ago
AI For Humans
16.5k views • 1 month ago
AI Explained
92.6k views • 1 month ago
AI Explained
56.7k views • 1 month ago
AI For Humans
10.3k views • 1 month ago
AI Jason
221.8k views • 1 month ago
AI Explained
72.1k views • 1 month ago
AI For Humans
11.3k views • 1 month ago
AI For Humans
3.4k views • 2 months ago
AI Explained
108.0k views • 2 months ago
AI For Humans
14.8k views • 2 months ago
AI Jason
97.6k views • 2 months ago
AI Explained
135.6k views • 2 months ago
AI Explained
93.2k views • 2 months ago
AI For Humans
7.6k views • 2 months ago
AI Jason
211.8k views • 2 months ago
AI For Humans
949 views • 2 months ago
AI Jason
16.2k views • 2 months ago
AI For Humans
987 views • 2 months ago
AI For Humans
1.1k views • 2 months ago
AI Explained
117.1k views • 2 months ago
AI For Humans
11.2k views • 2 months ago
AI Jason
82.4k views • 2 months ago
AI For Humans
1.0k views • 2 months ago
AI Explained
109.7k views • 3 months ago
Andrej Karpathy
1.4M views • 3 months ago
AI Explained
135.2k views • 3 months ago
AI Jason
189.7k views • 3 months ago
AI Explained
111.2k views • 3 months ago
Andrej Karpathy
2.6M views • 3 months ago
AI Jason
14.7k views • 3 months ago
AI Explained
122.8k views • 3 months ago
AI Jason
18.1k views • 3 months ago
AI Explained
107.6k views • 4 months ago
AI Explained
182.7k views • 4 months ago
AI Jason
51.4k views • 4 months ago
AI Explained
106.0k views • 4 months ago
AI Jason
46.6k views • 4 months ago
AI Jason
43.4k views • 4 months ago
AI Explained
108.3k views • 4 months ago
AI Jason
68.1k views • 4 months ago
AI Explained
287.4k views • 5 months ago
Andrej Karpathy
34.7k views • 5 months ago
AI Jason
96.7k views • 5 months ago
AI Explained
87.3k views • 5 months ago
AI Explained
74.9k views • 5 months ago
AI Explained
153.6k views • 5 months ago
AI Explained
116.9k views • 5 months ago
AI Jason
62.6k views • 6 months ago
AI Jason
325.6k views • 7 months ago
AI Jason
151.4k views • 8 months ago
AI Jason
201.3k views • 8 months ago
AI Jason
31.3k views • 8 months ago
AI Jason
19.7k views • 9 months ago
AI Jason
124.6k views • 10 months ago
Andrej Karpathy
811.3k views • 11 months ago
Morningside AI
13.8k views • 1 year ago
Andrej Karpathy
813.6k views • 1 year ago
Morningside AI
4.2k views • 1 year ago
Morningside AI
10.1k views • 1 year ago
Andrej Karpathy
2.8M views • 1 year ago
Morningside AI
26.7k views • 1 year ago
Andrej Karpathy
5.7M views • 2 years ago
Andrej Karpathy
223.2k views • 2 years ago
Andrej Karpathy
268.3k views • 2 years ago
Andrej Karpathy
384.0k views • 2 years ago
Andrej Karpathy
423.4k views • 2 years ago
46 Comments
we are less than 1% into the development of AI. There is no clear winner at all. It's a development, not a race with a goal line that will eliminate everybody else.
That's ridiculous. I have tested the new Opus yesterday and it's way behind O3. Not only in terms of reasoning but also tool usage. In fact Qwen3 32b did clearly better in my tests.     See More
From my experience once the chat gets longer starts to loose the smartness. I tried it in API to develop three.js scenes, follow-up prompts will immediately break the working parts of the co     See More
Claude 4 will report people to the police if it thinks you do something bad. I think this is the most dangerous thing about this model
Google releases new "Gemini Perfected Ultra Instinct" model.
Anthropic: "Oh yeah? Well our new Claude 5 Magnum Opus can cut a bitch."
The channel “AI Explained” explained that the new Claude model did NOT trigger the ASL-3. They had apparently already planned to do ASL-3 prior to even starting training Claude 4. Probab     See More
Blue Player 3 found a bug — he’s scoring on his own goal (red got 172 points), getting exp for his level but at the cost of his team winning.
Also, red was trying to do that too,     See More
did the title say Claude 3 this whole time? 🤦     See Less
we are less than 1% into the development of AI. There is no clear winner at all. It's a development, not a race with a goal line that will eliminate everybody else.     See Less
TIMESTAMPS & Summary (by VidSkipper     See More Claude 4 Opus is announced as a powerful AI model triggering stricter safety measures due to its potential misuse, while also showcasing impressive coding and problem-solving capabilities across various applications like 3D environment generation and game development.
0:00 🚀 Claude 4 Opus Announcement
• 🚀 Anthropic announced Claude Opus 4 and Claude Sonnet 4, surpassing previous models like Sonnet 3.7, OpenAI's CodeX 1, and Gemini 2.5 Pro on the SUI benchmark.
• 🛡️ Claude Opus 4 triggers stricter AI safety measures due to potential misuse in developing chemical, biological, radiological, and nuclear weapons, leading to its classification under AI Safety Level 3 protections.
• 📊 Anthropic's models are categorized under ASL1-4, with Claude Opus 4 being the first to reach ASL 3, indicating significantly higher risk potential compared to DeepMind and OpenAI's previous models.
4:52 💰 Pricing and Features of Claude 4
• 💸 Claude Opus 4 and Sonnet are available with various plans, including a $200/month option, aligning with Google and OpenAI's pricing.
• ⚙️ Users can enable 'extended thinking mode' for Claude to think out loud for longer, addressing trickier questions and challenges.
• 🎨 Cloud 4 Opus with extended thinking can generate intricate 3D environments like a Minecraft castle from a few prompts, showcasing its advanced coding and rendering capabilities.
8:39 🧪 Coding and Simulation Abilities
• 🎯 Claude Opus 4 created a solar system simulation where players launch probes using planetary gravity, though tracking probe functionality was not successfully implemented.
• ✨ The model attempted to simulate the three-body problem, rendering a view of space and a view from a planet's surface, but the simulation had some collision issues.
• ⚽ A 2D soccer game was developed with players having stats and XP, but a glitch allowed one player to gain infinite XP and become unstoppable.
14:00 🌱 Updates and New Capabilities
• 📚 Anthropic is pushing updates with extending thinking and tool use, enhancing parallel task execution and memory capabilities, demonstrated by Claude playing Pokemon.
• 📝 Claude Opus 4 records key gameplay information to improve its performance, creating its own documentation, surpassing previous models in documentation quality.
• 💻 Claude Code is generally available with beta extensions for VS Code, integrating Claude 4 for coding assistance, similar to creating a personal coding assistant.
** Generated using ✨ VidSkipper AI Chrome Extension    See Less
That's ridiculous. I have tested the new Opus yesterday and it's way behind O3. Not only in terms of reasoning but also tool usage. In fact Qwen3 32b did clearly better in my tests.     See More 9;t test it's coding capabilities yet but the new Sonnet more or less instantly started hallucinating when given a task related to a somewhat poorly documented software. My impression is that Opus is more a or less screwed up release, just like the latest Llama 4. Training an AI is just like growing a sensitive plant: Sometimes it just doesn't work and nobody knows what went wrong.    See Less
From my experience once the chat gets longer starts to loose the smartness. I tried it in API to develop three.js scenes, follow-up prompts will immediately break the working parts of the co     See More e chat go longer! At least in my experience, but still overall is very great for short chats.    See Less
Claude 4 will report people to the police if it thinks you do something bad. I think this is the most dangerous thing about this model     See Less
Dangerous for your wallet     See Less
Google releases new "Gemini Perfected Ultra Instinct" model.
Anthropic: "Oh yeah? Well our new Claude 5 Magnum Opus can cut a bitch."     See Less
The channel “AI Explained” explained that the new Claude model did NOT trigger the ASL-3. They had apparently already planned to do ASL-3 prior to even starting training Claude 4. Probab     See More reasons…    See Less
Blue Player 3 found a bug — he’s scoring on his own goal (red got 172 points), getting exp for his level but at the cost of his team winning.
Also, red was trying to do that too,     See More layer 1 kept stealing the ball and passing back to Blue Player 3, so they were both getting skills at the cost of losing the game.    See Less