| Mobile Web

Search results for Qwen3

AI & Enterprise

Stanford, Meta develop AI inference technique to boost speed and cut costs

A research team from Stanford University, Together AI, the University of Illinois Urbana-Champaign and Meta introduced a new AI inference technique that processes intermediate reasoning in an internal latent space rather than generating text tokens. The approach combines latent and text modes and uses a HybridLink to keep reasoning continuous. In tests on Qwen3-based models, it reduced token generation and increased inference speed, while improving problem-solving performance. The work has not been peer-reviewed.

AI & Enterprise

Microsoft joins decision-focused AI model race with Microsoft-Decision-1

Microsoft has unveiled Microsoft-Decision-1, an AI model designed for fast decision-making that outputs structured results software can use directly. The model handles yes-or-no, multiple-choice and scoring requests and provides probability scores for each option. Microsoft trained it on the open-source Qwen3.5-9B model and plans versions based on its MAI model and OpenAI models. It cited internal use cases showing faster performance and lower costs versus GPT models.

AI & Enterprise

Major tech firms join race for decision AI models after Jev debut

The market for decision-focused AI models has shifted into a multi-player race after TypesafeAI, founded by former OpenAI employee Diogo Almeida, unveiled Jev in September. Unlike text-generating large language models, Jev returns decisions and probabilities from preset choices. OpenAI introduced a limited Decisions API, while AWS open-sourced Strands Decider 2B. Cloudflare launched Clef and Clef Flash, trained on Qwen models, and priced Clef at $0.24 per 1 million tokens.