Cohere’s Aya Vision 8B and 32B AI models excel in multilingual vision-language tasks, using synthetic data, model merging, and Pixel Shuffle compression. Open weights, benchmarks, and WhatsApp integration enable global applications.
OpenAI released GPT-4.5 on February 27, 2025, emphasizing improved usability and accuracy over GPT-4, though it falls short of frontier model status. Integration with Azure AI Foundry and a GPT-5 roadmap signal broader AI advancements.
Amazon introduces Alexa+, an AI-enhanced voice assistant providing personalized interactions, expanded functionalities, and integration with various services.
Google has launched Gemini Code Assist for Individuals, a free AI coding assistant with 180,000 monthly code completions and 240 daily chat requests, alongside a GitHub-integrated code review tool.
Anthropic released Claude 3.7 Sonnet, a hybrid reasoning AI model enhancing coding, math, and problem-solving. Priced competitively, it integrates with major cloud platforms and offers developer tools like Claude Code.
Google’s AI co-scientist aims to expedite scientific discoveries by collaborating with researchers, offering innovative hypotheses and streamlining experimental design.
xAI launched the Grok 3 model family on February 17, 2025, with a mini variant for speed and beta features like DeepSearch. Claims benchmark wins over GPT-4o and Claude 3.5.
Perplexity AI launches a free Deep Research tool, delivering comprehensive reports in under three minutes, democratizing expert-level analysis.
OpenAI has canceled the standalone release of its o3 AI model, choosing to integrate its technologies into the upcoming GPT-5. This strategic move aims to simplify product offerings and provide a more unified AI experience for users.
