Nuked 
Hello everyone! It’s a fantastic day to dive into the exciting world of AI.
AI models are popping up everywhere, with contributions from both big players and innovative startups alike. It’s easy to feel overwhelmed with so many options available.
TechCrunch has done the legwork for you by gathering insights on the most advanced AI models launched since 2024, detailing their functions and applications.
As a quick note, keep in mind that there are over a million AI models to choose from. Hugging Face, for instance, boasts over 1.4 million models!
In 2025, we saw the introduction of OpenAI’s GPT 4.5 ‘Orion’, acclaimed for its extensive knowledge and emotional intelligence. However, it hasn’t quite hit the benchmarks of its competitors in reasoning tasks.
Another noteworthy model is Claude Sonnet 3.7 from Anthropic. This hybrid model offers both speedy responses and deep thinking capabilities. Users can control how long it takes to think things through.
xAI’s Grok 3 has also made waves this year, particularly in math and coding, following a promise to enhance its political neutrality after previous iterations displayed biases.
Meanwhile, OpenAI has rolled out other models such as o3-mini, designed for STEM tasks, and the powerful Deep Research agent focused on thorough research outputs.
Mistral’s Le Chat AI assistant claims to be the quickest of its kind, responding more rapidly than its peers, even if it falls short sometimes on accuracy compared to well-known names like ChatGPT.
Looking back to 2024, we saw DeepSeek R1 make a splash in Silicon Valley for its impressive capabilities balancing coding proficiencies with an open-source model.
With options varying from free models to premium subscription-based versions, exploring the latest in AI technology has never been more accessible, albeit slightly daunting!
