OpenAI Pando, a Git-compatible monorepo backend, cut median push latency nearly 80% and serves nearly 1 PB daily.

Zyphra Hierarchical Attention (ZHA) retrofits pretrained LLMs with a tree indexer, adding 81,920 parameters, nearly matching dense accuracy.

Deep Principle released Hydrogen-1.0, a 753B open-weight scientific model post-trained from GLM-5.3, scoring 38.3% on FrontierScience-Research ...

Letter’s Letterpress-1 coding agent turns growth tasks into pull requests, offering 760 tools and pricing from $100 monthly.

There is no tool for browsing raw schemas. NVIDIA chose that on purpose, so agents reach data through the semantic layer. The ...

Vercel Labs agent-desktop gives AI agents computer use across macOS, Windows, Linux with 60 accessibility-tree commands, not ...

It is the second 0.6B release in Edge0’s Audio8 TTS family. It builds on Audio8-TTS-Preview-0.6b and keeps the same ...

LEANN is a low-storage approximate nearest neighbor (ANN) index built on top of FAISS. It targets one problem: vector indexes ...

NVIDIA Releases Nemotron 3.5 Transcribe, an open speech LLM with 4.3B active parameters and a 4.71% English WER.

Factory Droid 1.0 agent harness makes programmatic tool calling default, scoring 65.2% on Terminal-Bench 4 with Opus 5.5.

TokenGate adds learned token pruning to existing LLMs, cutting long-context prefill time up to 11.3x and memory 3.4x.

Liquid AI LFM3 ships 6 open-weight 2B-class models; LFM3-1.7B averages 49.37 across 29 self-reported benchmarks, above ...