Your laptop wasn't built for coding harnesses — and the numbers prove it
Nine coding harnesses vs. your laptop
A hands-on benchmark of nine coding harnesses running Qwen 3.8 27B locally on an M4 MacBook Pro reveals how poorly most tools handle localhost. Large system prompts and tool schemas can cost over three minutes of prefill before the first token, while side requests overlap agent turns and crush throughput. Lean harnesses like pi and mini-swe-agent stay fast with 96–99% cache reuse. The author's own chad, coupled in-process with MLX, jumps from 7.9 to 17.4 tokens per second.
It’s not your fault: most coding harnesses weren’t built with a local model in mind.
- OleksandrC
If you're looking for a coding agent that would fit nicely into resource-constrained environments (such as laptops, or tiny VPS servers, or tiny single-board computers, etc), and would also work great with local models - you might also like hax (https://usehax.dev/). 0.7 MB dynamically linked native C binary, few MBs of RAM usage when running, auto-discovers config from running local llama-server, and uses minimalist system prompt and tools for lean context usage.
- julesrms
HN seems to have had a stream of agent harness benchmarks floating past. And every time I wonder where the people who create these tests are looking when they're deciding which harnesses to test? Because right now nobody seems to bother testing mine! (https://juggler.studio)
I know Juggler's very new, but there's so much churn going on in this area that it's hard to know where I should be pushing it. It's hard to guess whether juggler's strengths would played well with a particular test like this, or made it look bad, all feedback about the kind of parameters people are interested in is useful to know when I'm deciding what to optimise.
- alex_john_m
What is this supposed to mean?
"it spreads up to 50% between nights, so nothing between the lean arms is a finding."
- toasty228
A bit off topic because I'm not using local models, but I recently benchmarked codex vs pi vs omp with my workload and found codex to be both faster and more token efficient than pi/omp. There was not a single case for which pi was faster/cheaper
- larodi
I can see this pattern of many people using Qwen 3.8 27B for local inference both on Apple Silicon and x86. This implies the model must be very good, given all these peoples' opinion converges on it.