Measured results and explainers from Glyd, lossless compression for AI model weights.https://getglyd.com/https://getglyd.com/reports/dfloat11-zipserv-rtx-4080-super/https://getglyd.com/reports/dfloat11-zipserv-rtx-4080-super/Qwen3-8B on a 16 GB RTX 4080 SUPER: Glyd generates 2.5 to 3.4 times DFloat11's tokens a second at the same size, and its kernels keep pace with ZipServ's.Sat, 26 Sep 2026 00:00:00 GMTReporthttps://getglyd.com/reports/qwen3-32b-one-48gb-gpu/https://getglyd.com/reports/qwen3-32b-one-48gb-gpu/Glyd holds Qwen3-32B's weights in 44.5 GB instead of 65.5, so it runs on one 48 GB RTX A6000 instead of two, every weight exact and faster.Sat, 26 Sep 2026 00:00:00 GMTReporthttps://getglyd.com/reports/ten-open-models/https://getglyd.com/reports/ten-open-models/Every Linear layer's matrix of ten popular open models, from SmolLM3 3B to Llama 3.3 70B, packed and unpacked bit for bit: 31.9 to 33.0% smaller.Sat, 26 Sep 2026 00:00:00 GMTReporthttps://getglyd.com/blog/bf16-exponent-2-6-bits/https://getglyd.com/blog/bf16-exponent-2-6-bits/A bf16 weight spends 8 of its 16 bits on the exponent, but in a trained model those 8 bits carry about 2.6 bits of information. Why, and what it buys.Sat, 26 Sep 2026 00:00:00 GMTExplainer