v5.2.0: GLM-5, Qwen3.5, Voxtral Realtime, VibeVoice Acoustic Tokenizer
The open-source AI landscape just got a massive power-up with three major model releases.
Deep Dive
Hugging Face Transformers v5.2.0 integrates three major new models. Zhipu AI's GLM-5 scales to 744B parameters (40B active) for complex agentic tasks. Qwen's 397B parameter Qwen3.5 model activates only 17B per pass for high efficiency. Mistral AI's VoxtralRealtime enables low-latency, streaming speech-to-text. The update delivers best-in-class open-source performance in reasoning, coding, and multimodal tasks, significantly closing the gap with proprietary frontier models.
Why It Matters
Developers now have access to state-of-the-art, efficient open models that rival closed-source alternatives in key benchmarks.