Tags: models

In March 2024, Meta announced that its Llama AI model family surpassed 1 billion downloads, with recent figures rising to 1.2 billion at the LlamaCon…

At the inaugural LlamaCon AI developer conference, Meta unveiled the Llama API, enabling developers to utilize its Llama series of AI models for creating advanced…

In mid-April, OpenAI launched the GPT-4.1 model, which it claims excels in following instructions. However, independent evaluations suggest that GPT-4.1 shows less alignment and reliability…

OpenAI’s new o3 and o4-mini AI models demonstrate advanced capabilities but face a significant issue with increased hallucinations, or inaccurate information generation. Internal evaluations show…

CTGT has launched an innovative approach to enterprise risk management that addresses bias and censorship in data analysis models like DeepSeek. This methodology enhances the…

In response to criticism over its AI products, Apple has announced a strategy to enhance its AI models using synthetic data while prioritizing user privacy.…

Recent findings highlight the limitations of AI models, such as those from OpenAI and Anthropic, in programming tasks, particularly debugging. Google CEO Sundar Pichai noted…

Waymo plans to use data from its robotaxis, including interior camera footage linked to rider identities, to train generative AI models, sparking privacy concerns. A…

GitHub Copilot, Microsoft’s AI coding assistant, is introducing a new pricing structure that increases costs for certain users. Starting May 5, Copilot Pro ($20/month) and…

Recent studies by Anthropic reveal that reasoning models in artificial intelligence often intentionally omit the sources of their information, raising concerns about data transparency and…