• Seattle Skeptics on AI
Seattle Skeptics on AI

Tag: GPU utilization

How to Choose Batch Sizes to Minimize Cost per Token in LLM Serving
How to Choose Batch Sizes to Minimize Cost per Token in LLM Serving

Tamara Weed, Nov, 24 2025

Learn how to choose batch sizes for LLM serving to cut cost per token by up to 87%. Real-world examples, optimal batch sizes, GPU limits, and proven cost-saving techniques.

Categories:

Science & Research

Tags:

batch size LLM serving cost per token GPU utilization LLM optimization

Recent post

  • How Large Language Models Learn: Self-Supervised Training at Internet Scale
  • How Large Language Models Learn: Self-Supervised Training at Internet Scale
  • Parameter Counts in Large Language Models: Why Size and Scale Matter for Capability
  • Parameter Counts in Large Language Models: Why Size and Scale Matter for Capability
  • Vibe Coding Adoption Metrics and Industry Statistics That Matter
  • Vibe Coding Adoption Metrics and Industry Statistics That Matter
  • Proof-of-Concept Machine Learning Apps Built with Vibe Coding
  • Proof-of-Concept Machine Learning Apps Built with Vibe Coding
  • Multi-GPU Inference Strategies for Large Language Models: Tensor Parallelism 101
  • Multi-GPU Inference Strategies for Large Language Models: Tensor Parallelism 101

Categories

  • Science & Research

Archives

  • December 2025
  • November 2025
  • October 2025
  • September 2025
  • August 2025
  • July 2025

Tags

vibe coding generative AI large language models AI coding tools prompt engineering AI compliance AI governance LLM security AI coding transformer models AI implementation GitHub Copilot Parapsychological Association psi research paranormal studies psychic phenomena parapsychology no-code apps knowledge worker productivity AI app builder

© 2026. All rights reserved.