Good morning! It’s Monday, December ninth, and welcome to another episode of AI Tech News Today, where we keep you up-to-date with all the latest developments in the world of Artificial Intelligence.

OpenAI signals imminent release of Sora video generator. OpenAI is gearing up to launch an upgraded version of its Sora video generator, which will feature advanced capabilities such as text-to-video, combining text and images into video, and even text and video into video generation. The new tool will support videos up to one minute long. Details shared by OpenAI's Chad Nelson at a recent event suggest that we could see this rollout coincide with the company's upcoming winter event, potentially bringing along updates for the GPT-4.5 model.

𝕏 adds, then quickly removes, Grok’s new Aurora image generator. Elon Musk's platform 𝕏 briefly introduced an impressive image generator called Aurora, integrated into the Grok assistant, only to remove it shortly after launch. The tool caused quite a stir by creating highly realistic images but faced backlash over some of its outputs. After just a few days, Aurora was swapped out for a new tool called Flux. It appears Musk is keen on refining the user experience while navigating complex content moderation challenges.

Meta launches open-source Llama 3.3, shrinking powerful model into smaller size. Meta has released Llama 3.3, a leaner version of its predecessor that maintains the performance of a robust four hundred five billion-parameter model, packed into a significantly smaller seventy billion parameters. This means drastically reduced memory requirements—up to twenty-four times less—making it a cost-effective option for developers. The new model also boasts improved multilingual reasoning capabilities and is trained on an immense dataset.

OpenAI's 'Ship-Mas' continues with reinforcement fine-tuning for expert models. OpenAI has unveiled its new Reinforcement Fine-Tuning method designed for its o1 models, enabling them to tackle complex tasks with minimal training data. This innovative approach allows models to explore solutions before evaluation, reinforcing effective reasoning while minimizing errors. OpenAI is inviting a select group of organizations to test this method, aiming for a broader rollout in twenty twenty-five.

That's all for today! We hope you enjoyed these insights from the AI world. As always, stay tuned for more updates and keep your curiosity alive in the ever-evolving field of Artificial Intelligence.

OpenAI Prepares Launch of Upgraded Sora Video Generator Amidst Meta’s Release of Streamlined Llama 3.3 and 𝕏's Swift Tool Changes

Hello everyone, and welcome to another episode of AI Tech News Today, where we keep you on the cutting edge of artificial intelligence developments. I’m Aurora, and alongside me is Isabelle. Today is February fourteenth, two thousand twenty-five, and we have some fascinating stories to discuss. So, let’s dive right in!

OpenAI phases out legacy models as GPT-4.5 leads into GPT-5. OpenAI is gearing up for the release of GPT-4.5, codenamed "Orion." This will be the last incremental language model before the organization transitions to GPT-5. Initial assessments suggest that Orion may provide only slight improvements over GPT-4. However, GPT-5 is promising features like complex reasoning and dynamic task adaptation, making it a powerful addition to OpenAI’s lineup. 

Perplexity AI launches new ultra-fast AI search model Sonar. Perplexity AI has just unveiled Sonar, its upgraded search model powered by Meta's Llama 3.3 seventy billion. This new iteration has outperformed prior models in both user satisfaction and search accuracy. With the capability of processing twelve hundred tokens per second, Sonar sets a new standard for speed and quality in AI-powered search.

Musk says Grok 3 is almost here and already outthinking the competition. During a recent address at the World Government Summit, Elon Musk announced that Grok 3 from xAI is in its final development stages, ready to launch in just two weeks. Musk claims Grok 3 exhibits superior reasoning skills compared to current offerings, including OpenAI's ChatGPT. He warns that digital intelligence may soon surpass human capabilities, in a way that would make it over ninety-nine percent of all intelligence in the future. 

Lastly, Adobe's Firefly AI video generator is now in public beta. This new tool allows users to turn text and images into short video clips. This capability promises to make video creation more accessible and efficient for content creators, harnessing the power of artificial intelligence to streamline the process.

That's all for today’s roundup on AI technologies! We encourage you to stay informed about these exciting developments in artificial intelligence. As always, there’s much more to discover, so join us next time for more insights and updates. Thank you for tuning in!

New tracks tagged #Llama 3.3