#huggingface

Yandex unveils AliceAI-Foundation‑80B‑A3B‑Base, a Russian‑crafted 80B LLM designed to rival Qwen 35B and DeepSeek V4 Flash with a custom architecture. Explore its specs, market impact and geopolitical context.

New XiaomiMiMo LLMs, like MiMo‑V2.6‑Flash‑RL and MiMo‑V2.6‑Distill‑Qwen‑9B, outperform rivals on Hugging Face with low VRAM and open‑weight access – test them now.

Qwen 3.8 and Qwen-Image‑2.1 lead the way in lightweight, high‑speed multimodal AI—benchmarks show up to 4,700 tok/s on GPUs and new image models support RGBA editing with 10‑image references.

Discover how a 2025 open‑source Jev architecture outshines frontier labs with RL‑based, non‑autoregressive design, JSON schema and full transparency.

Alibaba releases open‑source AI model DAMO‑RADAR for detecting cancer and 150+ conditions, sparking debate on accessibility, clinical impact and industry influence.

Baseten launches a new safety framework for open‑weight AI models, partnering with Hugging Face to evaluate and curb “abliteration” and its regulatory implications.

Open‑source Laya outperforms Jev: 421M‑parameter, 35 ms decision engine for real‑time gaming & web automation.

Discover how small‑scale language models like Needle 3 and Swift Qwen 3.8 are redefining AI: efficient, on‑device, and high‑performance with minimal parameters.

AI trends 2026: Chinese open‑weight models close the US gap, Hugging Face breach highlights supply‑chain risks, and GPT‑6 Astra excels in gaming. Get the latest insights now.

Discover Jev’s non‑autoregressive architecture—fast, cost‑effective structured outputs vs. classic LLMs. Explore its speed, challenges, and open‑source debate.