Hacker News

I'm really happy this is one of the top comments here, I am fully local as well.

Just wanted to leave a note for folks who might not have the memory to run a big 32gb model - I just found out there are some pruned models that have really good performance and If I had a smaller machine I might try this pruned unsloth Q4 quant of GLM 4.7 flash that sits at 14gb: https://huggingface.co/unsloth/GLM-4.7-Flash-REAP-23B-A3B-GG...

I usually use LM Studio for this type of thing but unsloth has their own studio type app that might be even better suited for these quants.

I used GLM 4.7 flash as my main model for months and it was an incredibly tenacious model and very very fast - I think on restricted hardware, this could be a great choice.

Qwen3.6-35B-A3B-UD-Q4_K_M runs at about 11 tokens/second on my poor old 1060. Absolutely nuts how far we've come

People want to make it seem like you need to always use the latest and greatest frontier models to be taken seriously as a developer.

You really don’t need them. After a certain point, bigger models give diminishing returns. If you can get 80% of the productivity gain with a free local model, use the local model. It will still be way faster than doing everything by hand, but you also don’t have to pay for tokens to a cloud provider and the tools won’t be ripped away from you on a whim.

This is the new attitude enlightened people should adopt. Reject the arms race.

This sounds very appealing. What size Mac mini would I need for that?

I never got into any of the AI models because it was clear local first was going to be more valueable, if they were to replace coding tasks.

I tried out a few models and ended up going with either Qwen3-Coder-Next (no think, just do) and Qwen3.6-35B (thinking, w/llamacpp token budget). Created a customized prompt that works fairly well to around ~60k tokens and then is a toss up on whether it's poisoned itself or I've directly steered it into the wrong. When it's clear that's happened, if it's important to continue, ask it to write a doc then start fresh.

I don't kno whow any one cold have witnessed the last 2 decades of American VC funded tech startups and tell themselves, "you know, this will be a reliable technolgy with no hidden problems".

Even a sober technical evaluation is just two steps:

1. You're proposing to build a app on a non-deterministic model.

2. That model is hosted behind a non-deterministic system (model alignment, model guardrails, system context subterfuge, cost/token pricing)

---

So you want to build your app and you think you're going to kep up with both #1 and #2?

Same here, I’ve removed my credit card from Copilot and won’t be renewing

What features/workflows have you added?

> they should have waited until the competition changed their prices first.

They did...

They're literally just passing on the costs https://platform.claude.com/docs/en/about-claude/pricing

Anthropic just provides a subscription - which Enterprise usually doesn't want you to use because everything you're submitting through that will be trained on / becomes part of their model.

So If you use it without explicit permission from your employer you may be committing a contract violation which can have serious consequences - up to jail time - as they can sue you for that.

i ran out of claude credits for the first time at work in months and had to fallback to copilot.

pleasantly surprised, claude's way ahead in tooling but the ability to designate what model your subagents use and having access to all models is a better feature than all of what claude offers combine atm.

The only limit on the amount of ai can consume in a month a work is dollars, so anything that helps with cost is the best model/harness for me.

It also did a better job at smart designating subagents itself where as claude often used higher cost models.

The price hike was insane. My $dayjob is moving away from Copilot and into Claude Code subscriptions. In parallel we are testing AWS bedrock and Deepinfra for open weight models in preparation for when CC inevitably stops being such a good deal and aligns with actual token cost. Fun times.

> I am a huge fan of Copilot CLI. It just feels so logical and low-friction to use compared to Claude Code.

Honest question, can you ellaborate? If given the option, I use OpenCode but what do you find in Copilot CLI that makes you prefer it to Claude Code?

Letting them automatically pick the model is no longer sustainable, but there are some very efficient models that are capable of executing the plan created by a much nicer model. It’s kind of embarrassing to think that Microsoft’s auto model selection was choosing cutting edge reasoning models for tasks like resolving dependency conflicts back when their pricing was at a loss.

The harness is super important, what tools are available and the system prompts vary from harness to harness.

Anthropic seems to have a modest lead on their harness and models, so it’s a best-of-both-worlds scenario.

> I'm not sure what Microsoft is doing behind the scenes

It’s probably the exact same model, but the tools and the prompts around it are worse, so you get worse results.

I had a similar experience moving away from Copilot within Zed. Now using the reasonix harness for Deepseek that makes cache hits almost free. And that's with unsubsidized American providers like Digital Ocean or Cloudflare.

Same ,I switched to cursor. I told it how to invoke msbuild and it can edit away without needing a native Visual studio plugin.. no problems at all. Target language c++

GitHub Copilot costs have ballooned in recent week, what once took $100 requires $300. I like using Claude with VS Code through Copilot and I feel it’s given me much better code, that I can control the quality. It’s much more transparent than Claude Code. It’s open source but and the IDE interface gives so many more features to have you context and control over whats generated. The increase in cost isn’t purely due to their price increases but also the Opus models agents use more tokens. So I’ve moved to Claude Code and I’m happily still using Opus 4.6. Fable and 4.7 seem to do much larger units of work, go off on tangents and make assumptions that frequently results in slop.

My copilot quota finished in maybe 2-3 prompts with claude 4.8 opus. i was expecting it to suck but not this bad. it was good while it lasted though

Microsoft needs to offer cheaper option since they change to token based billing. GPT-5.4 used to be x1 for yearly subscriber but now it cost 6x. i run out the premium request for just couple prompts. Github copilot for $10 used to be the best value since you get all the US AI labs model for cheap.

> The performance, if we trust the benchmarks, put it at Sonnet 4.6.

I don't trust these benchmarks. I used a number of times Kimi K2.7 and I was disappointed. It would run in circles for things that Claude would do in one shot. However, my usage was via Ollama cloud, and I have no idea if they serve the actual model or a quantized version, and it was the quantization that degraded the performance.

The great news, in my opinion, is the precedent. If Microsoft is now serving Kimi K2.7, then very soon they might start serving GLM 5.2, and that is indeed a very competitive model.

> People have been asking for a way to run the Chinese models from a trusted provider

I'm going to be called a chiller again, but at this point I don't care as it is relevant. Synthetic runs their own models for a reasonable price, GLM5.2 & Kimi K2.7-Code included.

Referral link :

https://synthetic.new/?referral=kwjqga9QYoUgpZV

Btw if anyone is wondering, GPT 5.5 does the same garbage as 5.4 mini for 4 times the cost.

Fable manages to make a reasonable game, at a cost of 40 cents.

  X X O O O
  O O X X X
  X X X O O
  X O O X O
  X O X X O

Nice idea. I just asked Haiku to do the same in Claude Chat on iOS: it created a interactive react game, implemented the rules and let it play. Clever move for 1$ input and 5$ output, Anthropic!

While LLM models are bad at games, they are perfectly capable of writing a RL agent to train on the game itself.

when i will be extremely bored, I think I will make two models play chess against each other. I bet there's a chess benchmark / llm tournament already somewhere

The rugpull with the pricing change without further notice was not taken kindly by enterprice.

We just cancelled everyone's plans and rolled liteLLM out internally. We kept it for the insanely cheap tokens, but now that they've switched to the new pricing, they're just like openrouter, just with far fewer models.

They were, until they decided to commit suicide for the service.

For some reasons compilot seems dummer than vscode Claude or vscode codex. I can’t tell what’s the exact reason but it didn’t feel right

Their harness is terrible compared to any of the other cli based harnesses I test against. Like shockingly bad.

This comes up all the time at work because the vendor management people don’t understand the llm ecosystem and think Claude through copilot is the same as Claude through Claude code.

A simple side by side comparison will show dramatic under performance 3 or 4 times out of five when I’m asked to explain the difference.

Enterprises still have big contracts with github, those companies are imposing tight spending limits now and if the open weight models enable those limits to last a bit longer that's probably quite popular.

The V4 models are already in the Azure AI foundry so maybe a good chance of it coming.

https://docs.github.com/en/copilot/reference/ai-models/model...

Says that they are run by Moonshot

Legacy corps on the Microsoft steamboat.

It does, but it's very poorly documented and quite unstable (on purpose i think). What the other commenter said about the VSCode BYOK seems to be the more reliable way.

I tried adding a Foundry LLM as Github Copilot custom model and failed miserably. But with VSCode BYOK (and Github Copilot as the interfact) i did get it working, and i can now use Deepseek V4 Flash with Copilot.

AFAIK you can already use custom models in VSCode Copilot, but probably not for cloud workloads yet.

It has supported custom, local, any BYOM for quite a while.

I work at GitHub but even then I often use OpenRouter models in the CLI and Copilot App

Copilot Chat supports BYOK since Oct 2025 for the VSCode plugin: https://code.visualstudio.com/blogs/2026/06/18/byok-vscode

Azure. It was already available on the Azure AI Foundry before.

https://docs.github.com/en/copilot/reference/ai-models/model...

On servers that are subject to the CLOUD Act. Expect no GDPR compliance.

https://docs.github.com/en/copilot/reference/ai-models/model...

They are run by Moonshot itself, so probably china

where does it say that? its not available to me (also annual) at the moment via cloud but it said it is rolling out gradually, so I'm not too concerned. Tho I'm not overly excited either given Copilot pricing now; I reckon this should be at most 1x.