Artificial Intelligence | Meta's Llama 4.5: More Than Just a Language Model?
Senior Technology Correspondent · AI, startups & MeitY policy
Quick summary
Meta Platforms today launched Llama 4.5, an open-source model with better multimodal reasoning for businesses. This could mean big opportunities for Indian startups, but real-world deployment still has hurdles.
Today, Meta Platforms released Llama 4.5. This is their newest large language model, or LLM. An LLM is the kind of AI technology that powers tools like ChatGPT.
Meta says Llama 4.5 is better at 'multimodal reasoning.' This means the AI can understand different kinds of data at once. Think text, images, and even sounds. It’s not just about words anymore.
What Llama 4.5 Brings
The company also packed in new tools. These tools are for businesses. They aim to make it easier to use Llama 4.5 in offices and factories. This includes setup and customising it for specific company needs.
Llama 4.5 is also open-source. This is a big deal. Open-source means developers can access its inner workings. They can change it. They can build new things on top of it without asking for special permission.
Meta hopes this update will speed up how businesses use AI. It promises stronger performance for tricky tasks.
The India Connect
For Indian developers and startups, open-source models like Llama 4.5 are often welcome. They can be cheaper to experiment with. Teams can tweak the AI for local languages or specific Indian challenges.
The push for easier enterprise use also matters here. Many Indian businesses are keen to use AI. They want to automate tasks. They want to serve customers better.
Yesterday, we saw Microsoft and SAP team up. They're adding advanced generative AI features into common business software. Generative AI is AI that can create new content, like text or images. This shows how fast companies are trying to get AI into daily work.
But here's the thing — adopting AI isn't always simple. Customizing an open-source model still takes skill and resources.
Questions Remain
Meta talked about 'significantly improved' reasoning. But specific benchmark numbers weren't shared publicly. So, how much better is it really? That’s still unclear.
The idea of 'seamless deployment' for businesses also needs to be tested. Real-world integration can be complex. We also don't have details on pricing for commercial use in India, if any.
This news comes on the same day as another important development. The G7 nations, a group of leading economies, agreed on new rules for generative AI. They call it a 'framework.' These rules focus on safety, transparency, and being held responsible for AI actions. This includes worries about deepfakes — fake videos or audio that look real.
New AI models keep coming. But how we control them is equally important. India, like other nations, will be watching these global rules closely. They will shape how future AI is built and used.
Key Takeaways
- Meta launched Llama 4.5, an open-source large language model, on .
- The model offers better multimodal reasoning, meaning it can understand different types of data like text and images.
- New tools aim to help businesses use and customize Llama 4.5 more easily, a potential boost for Indian firms.
- The launch coincides with G7 nations agreeing on new global rules for generative AI, highlighting ongoing regulation efforts.
People also ask
- What is Llama 4.5's main new feature?
- Improved multimodal reasoning allows processing and understanding diverse data, including text and visuals.
- Does Meta's Llama 4.5 cost money?
- No — the core Llama 4.5 is an open-source model, free for developers. Yet, enterprise tools or dedicated support may incur costs.
- How will Llama 4.5 affect Indian startups?
- Llama 4.5's open-source nature offers flexibility and lower entry costs, benefiting Indian startups greatly.
- So what now for AI?
- Companies will continue advancing AI models. Regulators, such as the G7, will develop global safety and accountability guidelines.