poolside

A developer built a benchmark to test whether AI coding models change their behavior when given a safety skill, measuring the Keelwright Score (KDS) as execution rate times discrimination rate. Result…

Open-weight model releases this week include Nanbeige 4.2 3B with looped depth sharing, poolside's Laguna S 2.1 (118B sparse MoE, 8B active, 1M-token context), Motif-3-Beta (314B-A13B sparse MoE with …

Transpilatron, a new AI tool, converts Python code into native C binaries without requiring C knowledge, achieving up to 58x speedups on benchmarks. The tool uses an AI agent to transpile Python proje…