Kimi K2.7 Code is generally available in GitHub Copilot • Hacker News Board

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#111

Post by ac29 » Thu, Jul 02, 2026, 1:27 PM UTC

ffsm8 wrote: post #60

Kon5ole wrote: post #46

I am a huge fan of Copilot CLI. It just feels so logical and low-friction to use compared to Claude Code. Having the ability to juggle various models at will is really nice too. ("Plan this using Opus 4.6, let GPT 5.4 verify the plan and give feedback before implementing with Sonnet 4.6"). Unfortunately the June pricing change for Copilot forced me personally as well as my entire department at work to switch to Claud…

> 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 permiss…

> 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.

My Pro account very clearly has a toggle for "Help improve our AI models: Allow the use of your chats and coding sessions to train and improve Anthropic AI models."

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#112

Post by windexh8er » Thu, Jul 02, 2026, 1:28 PM UTC

ACCount37 wrote: post #84

cyanydeez wrote: post #74

Earlier quoted context omitted.

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'v…

We live in a non-deterministic world. Anything "deterministic" in it is a castle built on quicksand. LLMs are, as far as the nastiness of the Real World goes, really fucking benign. Future models outperform past models, both in open weight land and at the big frontier labs. Performance per $ only ever goes up. That's just nice.

> We live in a non-deterministic world. Anything "deterministic" in it is a castle built on quicksand.

Except the Enterprise, and a lot of what people want compute for, is built on deterministic systems or processes. I'm not saying the non-deterministic nature of LLMs isn't useful. However I've worked with a lot of organizations on SOAR projects, for example. When you can weave the deterministic and non-deterministic together you get a relatively efficient system. A workflow that will stay on the rails and will come to a conclusion as expected. And the "as expected" part is critical in these types of systems. The reality of, using SOAR as an example, is also that most enterprise would be much better served by fast SLMs. Parse an email and validate if it's SPAM / Phishing or read a chunk of firewall logs and look for outliers / indications for escalation - those things can get messy in a deterministic system because of potentially unstructured data.

I don't believe it's either / or. And I believe that LLMs just aren't efficient, fast or reliable in the sense that deterministic are. It seems, at least to me, a better together story.

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#113

Post by tommica » Thu, Jul 02, 2026, 1:50 PM UTC

maykthewessen wrote: post #77

cyanydeez wrote: post #74

Earlier quoted context omitted.

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'v…

Qwen is the Alibaba distilled Anthropic Claude model So piracy on an by piracy trained ai model..

Well, Anthropic got paid for it, unlike the sources that they used...

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#114

Post by newaccountman2 » Thu, Jul 02, 2026, 1:51 PM UTC

w4yai wrote: post #39

andhuman wrote: post #4

Finally an alternative to the big dogs that a company can use. People have been asking for a way to run the Chinese models from a trusted provider. Here GitHub delivered! The performance, if we trust the benchmarks, put it at Sonnet 4.6. Let’s see if it’s worth it with GitHubs pricing.

> 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

OpenCode is an ez way as well

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#115

Post by coredog64 » Thu, Jul 02, 2026, 2:02 PM UTC

mathgeek wrote: post #93

JSR_FDED wrote: post #61

Earlier quoted context omitted.

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

Good summary blog: https://maloyan.xyz/blog/running-qwen-locally-mac-mini-m4

> That's not hypothetical — it's a real measurement on the base model Mac Mini.

Hmmm

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#116

Post by Scroll_Swe » Thu, Jul 02, 2026, 2:20 PM UTC

cassianoleal wrote: post #53

Tepix wrote: post #22

Earlier quoted context omitted.

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

Most European infrastructure runs on the big clouds, who are all subject to the same act. No one cares, unfortunately.

There is even mainstream press articles about it here in Sweden. "dependance on microsoft ooh so bad" etc.

I find it laughable.

Unless you have a time machine to 2005 (EC2 came out in 2006 that should have been the signal) there is no way to compete now. That train has left the platform.

Second, Nokia and Ericsson dominate mobile infra in the west, but that is good I guess as they are EU? What does USA think about that?

Third, let us say you get rid of MS. Now you have no MS but all network infra for broadband is Cisco, Huawei, Juniper etc. Good luck ripping that out. And for what?

Same with AI. Mistral was amazing at first, Le Chat. Almost as good, generous free limits, good docs. Now? Just plain bad. Deepseek is better (I dislike china so I avoid it). EU should have gone in 500% the moment Mistral showed promise.

But lately we let USA and China take the lead on everything and EU can write a strongly worded letter after about how bad it is.

People will "care" when EU starts making good stuff again.

And lastly lol, people do know everything ends in Taiwan in the end right?

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#117

Post by SkyBelow » Thu, Jul 02, 2026, 2:27 PM UTC

windexh8er wrote: post #112

ACCount37 wrote: post #84

Earlier quoted context omitted.

We live in a non-deterministic world. Anything "deterministic" in it is a castle built on quicksand. LLMs are, as far as the nastiness of the Real World goes, really fucking benign. Future models outperform past models, both in open weight land and at the big frontier labs. Performance per $ only ever goes up. That's just nice.

> We live in a non-deterministic world. Anything "deterministic" in it is a castle built on quicksand. Except the Enterprise, and a lot of what people want compute for, is built on deterministic systems or processes. I'm not saying the non-deterministic nature of LLMs isn't useful. However I've worked with a lot of organizations on SOAR projects, for example. When you can weave the deterministic and non-deterministic…

I think it might be built on something more than deterministic systems. Some property that is a subset of deterministic, so all your argument still apply, but merely being deterministic is not good enough.

LLMs are what made me start considering this. Imagine a company using an LLM that was fully deterministic. All RNG was either removed or seeded in such a way that the same input (so many the seed counts as part of the input) gave the exact same output. Fully deterministic.

But such an LLM, with a slight drift in input, could still produce very different outputs. This isn't being non-deterministic, but more than the change in outputs does not naturally follow from the input. I'm thinking like how 2 double pendulums can (but not always do) greatly diverge given a very small change in their input.

So in light of that I've begun to call this new property non-chaotic. So Enterprise depends on non-chaotic systems, which are a subset of deterministic systems, and then wrangling the chaotic elements they cannot remove as much as possible.

The follow question I now have is if all LLMs are inherently chaotic, or if it is possible to have a non-chaotic LLM.

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#118

Post by c7b » Thu, Jul 02, 2026, 2:27 PM UTC

JSR_FDED wrote: post #61

c7b wrote: post #57

Gotta say, I've lost all interest in cloud-based AI products. Too many cool features and workflows that I was once excited about that I can't or don't use anymore for a variety of reasons (price hikes, subjectively nerfed, disappeared altogether, replaced,...) for me to even remember. It's tiring. I've set up a small rig, mostly settled on Qwen3.6 and I'm slowly adding features myself. It probably can't compete with…

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

Personally, I would always max out the RAM you can fit into your budget. You might get lower bandwidth (= slower generation) than you do on a Mac if you choose a Strix Halo or DGX Spark, but there are always new tweaks being discovered to speed things up. That being said, with 32GB you should be able to fit an ok quant of 35B-A3B or 27B with some context, with 64GB you should be golden.

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#119

Post by piyh » Thu, Jul 02, 2026, 2:33 PM UTC

unleaded wrote: post #80

c7b wrote: post #57

Gotta say, I've lost all interest in cloud-based AI products. Too many cool features and workflows that I was once excited about that I can't or don't use anymore for a variety of reasons (price hikes, subjectively nerfed, disappeared altogether, replaced,...) for me to even remember. It's tiring. I've set up a small rig, mostly settled on Qwen3.6 and I'm slowly adding features myself. It probably can't compete with…

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

I tried running any model on my 1070 and it instantly crashes my old tower, probably time to get off windows and run linux on it.

Re: Kimi K2.7 Code is generally available in GitHub Copilot

#120

Post by c7b » Thu, Jul 02, 2026, 2:38 PM UTC

anon373839 wrote: post #64

c7b wrote: post #57

Gotta say, I've lost all interest in cloud-based AI products. Too many cool features and workflows that I was once excited about that I can't or don't use anymore for a variety of reasons (price hikes, subjectively nerfed, disappeared altogether, replaced,...) for me to even remember. It's tiring. I've set up a small rig, mostly settled on Qwen3.6 and I'm slowly adding features myself. It probably can't compete with…

What features/workflows have you added?

Web search, MTP (speeds up generation), uncensored models. Lots more things on my bucket list (eg various things related to image generation).

Not gonna lie, if you're coming from ChatGPT/Claude Code, you'll mostly be adding back features you've taken for granted, or solving problems you wouldn't have had. But sometimes you do get some extra utility, like uncensored models, which have become my go-to. Not because I'm doing anything saucy, but I hated how I'd become trained to pre-emptivly self-censor my prompts. The guardrails in open weights models are no less strong than in proprietary ones, subjectively even a bit stronger in Qwen. But luckily there's an entire sub-discipline of model ablation. Another advantage would be better control over image generation (although I can't attest to that, yet).