Microsoft's Vibe Voice, Eleven Voice & More Crazy AI Updates!

0:00 Microsoft's Vibe voice copilot in Excel spreadsheet command a reasoning model Quen CLI 11 Labs music generation 11 Labs version 3 van 2.2 2 and much more to cover. That's exactly what we're going to see today. Let's get started. Microsoft released Vibe Voice 1.5 billion model, open-source texttospech model in preference, realism, and richness. This is better than 11 Labs version 3 and also Gemini 2.5 Pro preview text to speech. Here is an example. Last time I'm starting to think you just don't care. It's easier to say work was crazy than to just admit that I'm not a

0:45 priority for you anymore. That's not fair. Of course, you're a priority. So, you can see the expression there. Let's see the singing. Let me try to sing a part of it for you. It's been a long day without that is really nice. So, you can download this and it's available as open source and run it locally on your computer. Next, Grog 2.5 model is open source is the best model last year. Considering it's very large, it's not feasible to run it locally on your computer. Next, Level Labs voice to music flow in Level Lab Studio. Let's try this. You can now upload a video to 11 Labs and get unique music generated based on the context of the video. The music you're hearing now was generated

1:31 using our new video to music flow. [Music] And all these music are really nice. I'll put all the links in the description below for you to try that out. Next, we have Highfields AI product to video. So, you add a product. So, here we are adding a cake to the image and automatically the video you can see it converts to cake. Now we're adding a motorbike and the motorbike is there. Adding a car, the car is there. Adding glasses and the glasses are there. And you can try that on high fields.AI. Next we have co-pilot within Excel which means you don't need to write any

2:16 formulas from now on. AI can do this for you automatically. It can classify the feedback based on the comments. As you can see here, listing airport codes based on the provided country. And you can see the different airport codes and changing to Australia will give the different airport codes. That is really nice. Just upgrade your Excel and try that. Now, next we got command A reasoning model. Advanced model for enterprise reasoning toss in comparison with R1 GPOSS 120B and magistrol medium. This is better. As you can see here, here is a comparison without reasoning and with reasoning. With reasoning is clearly better. Deep research benchmark. This is in competition with Gemini 2.5

3:03 Pro. This is far better than Cloud Research, Perplexity Research and Gro Deep Search. Next, we got Quen CLI. Same like Claude code, Gemini CLI, we got Quen CLI, which you can try that. Now, considering the Quen models are open source, you can even run this locally on your computer and try this. You can also connect this with VS Code. Next, we got Quen image edit. This is open weights leader in image editing. Nano banana is good or better than this but still that is not open source but quen image edit is open source and you can see it's topping the list. Here are a few examples original and with edit we are adding the text here in the mug. Here is the original. Here is quen image edit compared with flux and gbt4. This is

3:49 really nice. Now we have prompt caching in gro which means it's 50% discount for cached tokens. Considering gro is already fast with many open source models. You can run this very cheaply. Next we have 11 music from 11 labs high quality AI music model. This is really nice. Let me try this. [Music] So you give a prompt and automatically all these music gets generated [Music] and you can see all the different types of music which gets generated.

4:38 Level Labs also released 11 version 3 alpha text to speech model. Hey, have you heard the big news? Hey, wait. Why are we whispering? I'm just trying to add some drama. Okay, because the V3 API is now available. Imagine all the possibilities. Exactly. Ladies and gentlemen, it's time to start building. This is really nice with dialogue mode, unlimited amount of speakers, 70 plus languages, enhanced voice and emotional control with audio tag. Here is 11 Labs V3 lip sync. The new 11 Labs verse 3 model is so good it's on fall now. What do you think? Oh, you can drop in your 11 Labs voice

5:24 ID to use a custom voice. It could turn these voices into a video, right? How do we do that? You can use the models on a that is really nice. Next in the same topic of lip syncing you got Skyre's A3 talking avatars. It is ranking better than Hydra for lip sync and Hon avatar. Here is an example. Hi Fum soggy soggy Mumbai. What's up tech fam? Today GPT5 is Toddwin with big buany steps. Claude goes peekaboo with mega context. This is from Sky Rails crafted with Grock and then with free pig and finally Sky reels. Next in the same topic for video generation you got van 2.2 S2V 40

6:09 billion parameter model designed for film grade audioddriven human animation. So you provide audio and image and it automatically lip sync for you. Let's try that. What people call love is just a chemical reaction that compels animals to breed. the first year. It is really nice considering it's open source, but the lip sync could be a little bit better. Next, cloud code has new context command to visualize your context window and token usage. Just type /context in your terminal and you can see the tokens used percentage, system prompt, system tools and various

6:54 other. You can optimize this further based on the stats. Next, we have Nvidia's Jetson Tour. The ultimate platform for physical AI and humanoid robotics. So, if you are interested in robotics, this would be really useful. Industry-leading performance for humanoid robots, supercomput for humanoids, high speeded sensor processing, robotic AI software, and robust security. Also, I came across a platform called Cance AI to generate video and image. Do try and let me know in the comments below what you think about this. I hope you like all these updates.