News
Voxtral 3 launches voice agent SDK with speech models in 50+ languages
Voxtral 3 provides multilingual speech‑to‑text and text‑to‑speech models, plus a SDK for real‑time listening, reasoning, and speaking, enabling voice agents on phone or web.
More News
- DeepInfer open-source repo enables offline inference on NVIDIA and Huawei hardware
- TorchTPU adds native PyTorch support for Google TPUs
- Analysis highlights AI bottlenecks in scientific discovery
- Report examines gap between GPU supply and compute demand
- Metis Reasoning proposes architecture to reduce hallucinations