English reports linked to this entity will appear here as their translations are published.
The Hugging Face blog announces NeoMME, a family of 260M and 800M-parameter multimodal and multilingual bidirectional encoders. NeoMME processes multilingual text and raw image patches in a single bidirectional Transformer encoder, replacing separately pretrained image and text encoders. All model checkpoints and a day-zero Hugging Face Transformers implementation are released under Apache 2.0.
Hugging Face published a blog post titled "Give Your Coding Agents a Memory You Own," introducing funes, a durable memory layer for coding agents. funes is built from sessions already on the user's machine and works locally, becoming part of the agent's normal workflow with one command. It supports coding agents including Claude Code, Codex, pi, and Hermes. The project lives at github.com/huggingface/funes.
IBM and Confluent announced that IBM Granite Time Series models are live in Early Access on Confluent Cloud, callable directly from Apache Flink via SQL functions for streaming forecasting and anomaly detection, with Confluent Platform to follow.
On September 1, 2026, Hugging Face released @huggingface/kernels, a JavaScript library for loading and running optimized WebGPU kernels from the Hub, along with an initial collection of 207 kernels and a new in-browser benchmarking tool called Fleet.
Ai2 published BenchMIRT on the Hugging Face blog, a new method for auditing LLM benchmarks at the level of individual prompts, designed to help researchers separate signals and identify what actually drives a benchmark's score.
Hugging Face announces that Indian-English joins the main Open ASR Leaderboard as Voice Arena Monsoon, in the default column set rather than as an opt-in toggle, so it contributes to the headline Average WER for every model. The blog also states that Monsoon is part of Voice Arena's broader dataset initiative for the Global South, and mentions public and private Hindi in the Multilingual tab.
Hugging Face launches an open community project to reproduce R1 and fill in DeepSeek's unpublished training data and complete training pipeline.
Hugging Face releases a small-parameter model family designed to run locally on hardware such as smartphones.
Hugging Face releases a dialogue model fine-tuned with direct preference optimization to improve conversational alignment.
BigCode releases an open large language model trained for code generation and programming tasks.
