The Acta blog

Ideas for better meetings

Practical articles on AI meeting assistants, meeting reports and turning conversations into decisions and work, from the team building Acta.

Cover: GGUF quantization explained, from Q8_0 to Q4_K_M and how to pick a file

GGUF Quantization Explained: Q4_K_M, Q5, Q8 and Which File to Pick

GGUF quantization in plain English: what Q4_K_M, Q5_K_M and Q8_0 mean, how size trades against quality, and how to pick the right file on Hugging Face.

Guides & How-tos 8 min read

Read article

Cover: how to read a Hugging Face model card, covering task tag, license and red flags

Guides & How-tos 9 min read

Hugging Face Model Card: How to Read One in 10 Minutes

How to read a Hugging Face model card: task tag, license, base model, card sections, GGUF files, the inference widget and red flags, with an example.

Read article

Cover reading LLM model size explained, with parameter sizes from 2B to 500B+ and the memory each needs

Guides & How-tos 10 min read

LLM Model Size Explained: What 2B, 9B, 27B and 500B+ Mean

LLM model size in plain terms: what 2B, 9B, 27B and 500B+ parameters mean, how much RAM each needs, active parameters, and how to pick one.

Read article

Cover reading mixture of experts explained, showing a router sending tokens to a few experts out of many

AI at Work 9 min read

Mixture of Experts Explained: Why Huge AI Models Run Fast

Mixture of experts in plain English: how routers pick experts, total vs active parameters in DeepSeek-V3 and Qwen3, and the hardware MoE really needs.

Read article

Cover: Ollama vs LM Studio vs llama.cpp vs vLLM compared for running open models

Guides & How-tos 8 min read

Ollama vs LM Studio vs llama.cpp vs vLLM: Which Local Runner Fits?

Ollama vs LM Studio, plus llama.cpp, vLLM, Jan and MLX: who each local LLM runner is for, GUI vs CLI, platforms, APIs, and a simple decision list.

Read article

Cover: choosing an open source LLM for meeting summaries, with context window, instruction following and model size

AI at Work 9 min read

Open Source LLM for Meeting Summaries: What to Try in 2026

Which open source LLM for meeting summaries holds up on long transcripts? Context length, languages, model size, and a test prompt you can run today.

Read article

Cover: open source LLM vs API for business, comparing self-hosting, APIs and finished tools

AI at Work 9 min read

Open Source LLM vs API for Business: Run It, Rent It, or Buy It

Open source LLM vs API for business: real cost drivers, privacy, quality and speed, with meeting notes as the example. When DIY wins and when it doesn't.

Read article

Cover: how to run a local LLM on a Mac or Windows laptop with Ollama or LM Studio

Guides & How-tos 8 min read

Run a Local LLM on Your Laptop: A Beginner's Guide (Mac and Windows)

How to run a local LLM on Mac or Windows with Ollama or LM Studio: hardware check, first prompt, a meeting summary task, and the privacy limits to know.

Read article