???? AI Accelerates: OpenAI, Google, Meta, and China Unleash a New Wave of Innovations

?? ChatGPT Now Creates Realistic Images and Google Launches AI Search in Spain ?? Meta Brings Its Assistant to WhatsApp and Manus, the Chinese AI, Challenges OpenAI

The pace of innovation in artificial intelligence shows no signs of slowing down. This week, five major announcements have once again shaken the sector and demonstrated that the battle to dominate the digital future is intensifying. OpenAI, Google, Meta, and the Chinese startup Monica have unveiled significant advances in image, voice, search, conversational interfaces, and reasoning models. The acceleration of technology is undeniable.

OpenAI adds realistic image generation to ChatGPT

On Tuesday, OpenAI introduced one of the most requested features to its flagship product: realistic image generation with GPT-4o directly inside ChatGPT. Until now, this feature relied on DALL·E, but it has now been integrated into the conversational experience, enhancing both coherence and accuracy. These hyperrealistic images are now directly competing with Elon Musk's Grok 3.

This new ChatGPT system can render up to 20 different objects with accurate relationships between colors, shapes, and attributes. It also significantly improves the model’s ability to generate text within images — a known weakness of visual generators up to this point. Users can refine images through multiple conversational iterations, and its autoregressive approach (generating the image step by step) allows for more detailed and controlled results. OpenAI states that all images include C2PA metadata to ensure traceability and prevent misuse. The feature is available for ChatGPT Plus, Pro, and Team users — and in a limited way for free-tier users.

Google: new AI-powered search and Gemini 2.5

The Mountain View giant has just launched “AI Overviews” in Spain, part of its search engine experience. This new feature, already available in the US and some European Union countries including Spain, offers a concise AI-generated summary at the top of the search results page when users ask complex questions. These answers are synthesized from reliable sources, and each segment includes a link to the original website.

The experience is being rolled out gradually and is only available for users over 18 who are logged into their Google account. While some web publishers have voiced concerns over reduced visibility, Google argues that it improves traffic quality and facilitates access to relevant information.

At the same time, Google has unveiled Gemini 2.5 Pro, its most advanced AI model to date. Already available to Gemini Advanced subscribers, the model clearly outperforms GPT-4.5 and o3-mini in several benchmark tests, such as Humanity’s Last Exam and various math and science exams.

Gemini 2.5 Pro can handle up to one million tokens (around 750,000 words) and is designed for multimodal tasks such as data analysis, code understanding, and solving complex problems using images, audio, and video. Google plans to expand this context window to two million tokens soon.

Meta: its AI assistant now speaks Spanish

Meta has also joined this week’s wave of announcements with the launch of Meta AI in Spain. After a long regulatory delay, Mark Zuckerberg’s company now offers its AI assistant on WhatsApp, Messenger, Instagram, and Facebook.

This assistant, recognizable by a blue circle icon within the apps, allows users to make queries, get web information, and generate content without leaving the app. Among its features is the ability to interact in WhatsApp groups using @MetaAI, as well as access contextualized answers, content suggestions, or personalized trends.

Meta AI is now available in six European languages (not including Catalan yet) and in over 60 countries. It has been used by more than 700 million people since its launch in the US. Meta plans to incorporate new creative functions and personalization enhancements in the coming months.

Manus: the Chinese AI challenging OpenAI

A new surprise from China has shaken the tech world: the startup Monica has introduced Manus, a general artificial intelligence agent that promises to execute complex tasks autonomously and, according to internal tests, outperform OpenAI’s Deep Research.

Manus is capable of programming, financial analysis, travel planning, and even real estate purchases without human supervision. Its creators claim that it “thinks and acts,” connecting ideas to actions. The agent has gone viral in China, even though it is only available by invitation. Access codes have reportedly sold for over €12,000.

However, Manus has also raised some concerns. Several experts report that it still makes errors in long and complex processes, and there is unease over its lack of transparency and the potential for collaboration with Chinese authorities for data collection.

Despite the criticism, Manus symbolizes a new generation of AI agents striving for true autonomy. Its growth is reminiscent of DeepSeek’s rise just a few months ago and has caught the attention of OpenAI and other major Western players.

In less than a week, five major developments have reshaped the landscape of artificial intelligence. From GPT-4o’s realistic image generation to Google’s AI-enhanced search, Meta’s assistant, and the autonomous Chinese agent Manus — all signs point to 2025 being marked by acceleration, format convergence, and a race to offer more for less. The user is the one who benefits most. AI is no longer a promise for the future — it is a rapidly expanding reality, and the tools launching today may radically change how we work, learn, communicate, and consume content. The revolution is being written in real time.