BENCHMARKS Function Calling Harness Pushes Qwen From 6.75 Percent to 100 Percent Success on Complex Schemas 8/10 4 min read 1 month ago
BENCHMARKS Liquid AI Runs 24-Billion-Parameter Model at 50 Tokens Per Second in a Web Browser 7/10 2 min read 1 month ago
TOOL UPDATES FlashAttention-4 Achieves 1,613 TFLOPs on NVIDIA Blackwell, 2.7x Faster Than Triton 7/10 4 min read 1 month ago
BENCHMARKS Humanity’s Last Exam Exposes the Gap Between AI Hype and Expert-Level Knowledge 8/10 2 min read 1 month ago
BENCHMARKS OpenUI Rewrites Rust WASM Parser in TypeScript, Achieves 3x Speed Gain 7/10 4 min read 2 months ago
BENCHMARKS Raspberry Pi 5 Runs Qwen3 30B at 7–8 Tokens/Sec with Potato OS 7/10 4 min read 2 months ago