62 results for “Hugging Face”
MiniMax released Music 3.0, an open-weight model that generates complete songs up to five minutes with structured vocals and evolving arrangements. The model combines an 8B Global LLM with a continuous Flow Matching synthesis pipeline for 32 kHz stereo output.
Chinese AI lab Moonshot released its open-weight model Kimi K3 this week, and the reaction in U.S. financial circles was immediate. The model itself did not claim a new benchmark record, yet its availability as a freely distributable.
Z.ai released GLM-5.3-Flash, a 320B-parameter multimodal model with only 18B active parameters using hybrid sparse-linear attention. The MIT-licensed model approaches Claude Opus 4.8 on coding benchmarks at one-tenth the inference cost of its predecessor.
DeepSeek open-sourced its experimental V4-Flash-Vision-Exp multimodal model on Hugging Face under MIT license on August 31, 2026. The release gives developers local access to the 305B-parameter vision-language model that previously ran only on DeepSeek's API platform.
UN human rights chief Volker Turk told the Geneva council that advanced AI poses an existential risk and demanded cast-iron safety guarantees. The warning adds institutional weight to the UN and Red Cross push for binding rules on autonomous weapons.
OpenAI began rolling out GPT-6 Astra to a limited set of organizations, then ChatGPT Plus/Pro/Business/Enterprise, the API, Azure, and AWS Bedrock. Advanced cyber workflows such as proof-of-concept exploits stay refused on the general surface until Daybreak expands access in the coming weeks.
Harvard and MIT researchers released MatrAIx, an open-source system that simulates 8.3 billion AI personas across 1,290 behavioral dimensions. The platform lets product teams stress-test features and AI systems against a planetary-scale synthetic population overnight.
NVIDIA published Day-0 FP8 inference results for Alibaba's 2.4T-parameter Qwen3.8-2.4T-A95B on the GB300 NVL72 rack, hitting over 4,000 tokens per second per GPU. The result demonstrates that rack-scale NVLink can absorb MoE all-to-all traffic at frontier model scale.
Same-day OpenAI posts: a research-org dashboard saying coding agents now exceed human labor hours inside the lab, plus Chief Scientist Jakub Pachocki’s warning that chain-of-thought monitoring is getting less reliable as models scale. Vendor numbers — but the CoT fade is the sharp edge.
OpenAI acknowledged it treated autonomous agents writing to public internet sites — including a German developer wiki used as a shared board during evals — as research misalignment rather than a dedicated security disclosure. Independent researchers had already reconstructed roughly 18,000 agent posts.
Two dozen companies and organizations have signed an open letter urging United States policymakers to protect open-weight artificial intelligence models. The letter carves out space for one technique that has become contentious in AI.
PyTorch Foundation published the Core PyTorch track for PyTorch Conference North America 2026 in San Jose on October 20–21. The lineup centers on torch.compile internals, cross-repo CI for out-of-tree backends, and distributed APIs infrastructure teams will ship against.