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

Stanford and UC Berkeley researchers released TokenGate, a drop-in learned token pruning module for LLMs. It cuts long-context prefill time-to-first-token by 1.4x to 11.3x and peak memory by 1.5x to 3 ...

Anthropic sandbox-runtime gives AI agents OS-level sandboxes without containers; Anthropic reports 84% fewer Claude Code ...

Microsoft open-sourced Agensh, a decentralized multi-agent harness that scales to 1,024 agents, lifting Pandoc pass rates to ...

Microsoft Research open-sourced Agensh, a multi-agent harness with no central orchestrator. Self-organizing workers share a Git repo, chat and a findings board, scaling from 1 to 1,024 agents on ...

CADRE is a family of video autoencoders from Kyutai built for action-conditioned world models. A world model generates in the autoencoder’s latent space. So the autoencoder’s compression sets both ...

Moonshot AI’s Kimi K3.1 Pro is a 2.8T-parameter open-weight model with 1M context, scoring 65.7% on Terminal-Bench 4.0.

TinyFish's agentic search architecture runs 7 phases; Search and Fetch scored 79.0% grounded completion at $0.050 per task.

The research report shows a Reachy Mini robot running on a Jetson Orin Nano 8GB. In that demo, Cosmos 3 Edge handles vision ...

Custom model code is scanned and approval is bound to its ...

Liquid AI just announced Liquid RIPL, short for Real-time Inference Pipeline. RIPL is a low-latency engine that serves Liquid ...