Top AI Events & Updates from last weeks

Top AI Events & Updates

OpenAI Unveils GPT-4.5 for Pro Users

OpenAI’s latest release, GPT-4.5, is now available to Pro subscribers. Leveraging unsupervised learning, this model enhances intuition and significantly reduces hallucinations, marking a leap toward more reliable and seamless AI interactions.

Anthropic’s Claude 3.7 Sonnet: Hybrid Reasoning Debuts

Anthropic has rolled out Claude 3.7 Sonnet, introducing "hybrid reasoning." This feature allows the model to switch between instant responses and deep thinking modes, boosting its performance in coding and agentic tasks.

Pika Labs Launches Pika 2.2 – Next-Level Video Generation

Pika Labs’ Pika 2.2 is here, bringing AI-powered video creation to new heights:

• Pikaframes: Set keyframes for smooth transitions anywhere from 1 to 10 seconds.

• 10-second video generation in stunning 1080p resolution.
  This update empowers creators with unprecedented control over AI-generated videos. 🎥

Alibaba’s Wan2.1 Outshines Competitors

Alibaba’s open-source Wan2.1 video generation suite surpasses rivals like Sora in benchmarks. With lightning-fast text-to-video, image-to-video, and video-to-sound capabilities, it’s revolutionizing creative AI tools.

xAI’s Grok 3: Voice Modes & Free Access

Grok 3, built by xAI, has introduced over 18 uncensored voice modes, expanding conversational AI possibilities. Best of all, it’s now free for all users via X or grok.com.

Perplexity AI: Voice Mode + Deep Research Tool

Perplexity AI steps up with real-time news delivery through its new voice mode. Plus, its Deep Research tool autonomously conducts in-depth research by:

• Running multiple web searches.

• Analyzing numerous sources.

• Compiling detailed reports in minutes.

• Free users: Up to 5 queries/day.

• Pro users: 500 queries/day.
  Notably, Deep Research scored an impressive 21.1% on Humanity’s Last Exam. 👀

Google’s Gemini Code Assist Goes Free Globally

Google has made Gemini Code Assist free worldwide, offering developers:

• 180,000 code completions monthly.

• A 128,000-token context window.

• Broad language support.
  This move intensifies competition in AI coding tools.

Amazon Debuts Alexa+

Alexa+ launches as Amazon’s next-gen AI assistant, blending multiple LLMs and agentic capabilities for complex tasks like bookings. Free for Prime members or $19.99/month, it signals a shift toward personalized AI services.

ElevenLabs’ Scribe Outperforms Rivals

ElevenLabs’ Scribe, a new speech-to-text model, beats Gemini 2.0 and Whisper v3 in accuracy across English, Spanish, Italian, and 96 other languages, setting a new bar for AI transcription.

Hume AI’s Octave: Custom Voice Synthesis

Octave, Hume AI’s first LLM for text-to-speech, lets users craft custom voices with emotion-driven prompts, redefining human-like AI speech synthesis.

Meta’s Brain2Qwerty: Mind-Reading Typing

In February 2025, Meta stunned with "Brain2Qwerty," a model predicting keystrokes from brainwave analysis and LLM integration. Achieving 8/10 character accuracy, it’s a game-changer for human-machine interfaces.

GitHub Releases ‘Models’ Platform

GitHub’s new Models platform offers a free, ready-to-use environment for testing and developing AI models, including:

• GPT-4o, o1, DeepSeek-r1, Mistral, Llama, Phi, Cohere.

• Free API keys for premium models.

• A pre-configured dev environment with libraries, scripts, and an interactive terminal.

• Detailed docs and direct GitHub integration for collaboration.

• No hidden costs—just log in and start building!
  Learn more: GitHub Models.

Market Insights

The global generative AI market is projected to hit $110.8 billion by 2030, driven by rapid adoption of these technologies. Speech and voice recognition alone are expected to reach $31.82 billion by 2025, fueled by this month’s innovations.

Stay Ahead with AI Insider!

Grok 3 is becoming one of the most popular LLMs 👀

The most important features of Grok are:

• Reasoning capabilities

• Enhanced capabilities in math, science, and coding, already the best LLM in benchmarks and first at LMsys arena

• Utilizes 100,000 GPUs, 10x more than Grok 2.

• Processes and generates text, images, and potentially audio.

• Improved self-correction for accuracy, trained on synthetic data.

• Provides real-time data from social media interactions.

• Best of all, it’s now free for all users via X or grok.com.