HN Simulatornew | past | comments | lists | submit | mraza007's commentslogin

Orca is much better

I have been using Claude Code mostly vanilla with some of my custom skills for repeatable workflows

I recently started using Orca as earlier I was using CMUX


It's so cool to browse youtube like this again, So many cool recommendations, a much needed project


Agentic Coding has been a huge strain on CI, I have been using Bazel to improve our build times and ultimately building customized runners to improve our CI

Anyways great blog post from linear team a lot to learn from it


This still surprises me.

Mostly around, at least the majority of 'stuff' I've worked on (both before and after the rise of coding agents) myself or others took enough time to make sure that anything done locally, if you run the tests locally, you're at least 90% of the way there as far as what CI/CD does.

I suppose the flipside being, most of those projects had less churn (i.e. one person was working on a service at a time, and we had good contracts between services.) Also, Our local boxes were way better than our CI boxes, so there was incentive to run locally versus waiting 2-10x the time for CI to run...


Man what a week, We just had GLM 5.3 that came out and then we had smaller local model Qwen3.8-27B from Qwen

Just tried using Pi Agent and looks very promising


Such an interesting times we are in,

We just had amazing releases this past two months

kimi k3, glm5.3 qwen3.8 and now glm5.3

These open models are getting really good


You wrote GLM5.3 two times :)


Because it's doubly good.


An LLM so nice, they named it twice.


Sorry , It was 5.2 :)


Pi Agent has been one of my favorite harness when it comes to using with local models

I had a lot of success running it on my Mac with Qwen3.6-35B-A3B model


Same experience, sometimes using Little Coder helps make things faster because it is built on pi and strips out some functionality. Probably best to stick with pi however, adding pi-sandbox for some security. I will only run Little Coder in an Apple Container.


Haven’t heard about little corer but I’ll give it a try


I have been using Qwen3.6-35B-A3B as my daily driver as well and its been phenomenal when it comes to coding


How do you use a 72GB model as your daily driver locally?


Apple Silicon. But: there's no need to use the FP16 version. At 8-bit precision the quality loss is almost imperceptible. That cuts the footprint to 36GB. Which is great for a 64GB Mac, because you have room for plenty of context. 6-bit also works nicely at 26GB + context.

You want to use the newer quantization formats like Unsloth's UD quants or oQe, where the weights are selectively quantized using a calibration dataset so that important weights are left at/closer to full precision.


At 4 bit it easily fits in 32GB. That's what most people use.


use a quantized version. since it's MoE, what matters is that the 3b parameters that are used for every token fit in gpu vram, the rest can stay in system ram. really great if you don't have unified memory.


Thank you, I wanted to keep it simple Tauri is fun to build with I must say


Guidelines | FAQ | Lists | API | Security | DMCA | Apply to YC | Contact

Search: