Get the top HN stories in your inbox every day.
danielhanchen
mathrawka
Looks like the docs have a typo:
Recommended context: 65,536 tokens (can be increased)
Adequate Output Length: We recommend using an output length of 65,536 tokens for most queries, which is adequate for instruct models.danielhanchen
Oh thanks - so the output can be any length you like - I'm actually also making 1 million context length GGUFs as well! https://huggingface.co/unsloth/Qwen3-Coder-480B-A35B-Instruc...
gnulinux
Do 2bit quantizations really work? All the ones I've seen/tried were completely broken even when 4bit+ quantizations worked perfectly. Even if it works for these extremely large models, is it really much better than using something slightly smaller on 4 or 5 bit quant?
danielhanchen
Oh the Unsloth dynamic ones are not 2bit at all - it's a mixture of 2, 3, 4, 5, 6 and sometimes 8bit.
Important layers are in 8bit, 6bit. Less important ones are left in 2bit! I talk more about it here: https://docs.unsloth.ai/basics/unsloth-dynamic-2.0-ggufs
blensor
Not an AI researcher here so this is probably common knowledge for people in this field, but I saw a video about the quantization recently and wondered exactly about that, if it's possible to compress a net by using more precision where it counts and less precision where it's not important. And also wondered how one would go about deciding which parts count and which don't
Great to know that this is already a thing and I assume model "compression" is going to be the next hot topic
CMCDragonkai
How do you decide which layers are the important ones?
PeterStuer
If you don't mind divulging, what resources and time did it take to dynamically quantize Qwen3-Coder?
tomdekan
Thanks Daniel. Do you recommend any resources showing the differences between different quantizations?
PeterStuer
I had given up long time ago on self hosted transformer models for coding because the SOTA was definetly in favor of SaaS. This might just give me another try.
Would llama.cpp support multiple (rtx 3090, no nvlink hw bridge) GPUs over PCIe4? (Rest of the machine is 32 CPU cores, 256GB RAM)
jychang
How fast you run this model will strongly depend on if you have DDR4 or DDR5 ram.
You will be mostly using 1 of your 3090s. The other one will be basically doing nothing. You CAN put the MoE weights on the 2nd 3090, but it's not going to speed up inference much, like <5% speedup. As in, if you lack a GPU, you'd be looking at <1 token/sec speeds depending on how fast your CPU does flops, and if you have a single 3090 you'd be doing 10tokens/ec, but with 2 3090s you'll still just be doing maybe 11tok/sec. These numbers are made up, but you get the idea.
Qwen3 Coder 480B is 261GB for IQ4_XS, 276GB for Q4_K_XL, so you'll be putting all the expert weights in RAM. That's why your RAM bandwidth is your limiting factor. I hope you're running off a workstation with dual cpus and 12 sticks of DDR5 RAM per CPU, which allows you to have 24 channel DDR5 RAM.
PeterStuer
1 CPU, DDR4 ram
danielhanchen
Oh yes llama.cpp's trick is it supports any hardware setup! It might be a bit slower, but it should function well!
ashvardanian
Thanks for the uploads! Was reading through the Unsloth docs for Qwen3-Coder before I found the HN thread :)
What would be a reasonable throughput level to expect from running 8-bit or 16-bit versions on 8x H200 DGX systems?
danielhanchen
Oh 8*H200 is nice - for llama.cpp definitely look at https://docs.unsloth.ai/basics/qwen3-coder-how-to-run-locall... - llama.cpp has a high throughput mode which should be helpful.
You should be able to get 40 to 50 tokens / s in the minimum. High throughput mode + a small draft model might get you 100 tokens / s generation
Abishek_Muthian
Thank you for your work, does the Qwen3-Coder offer significant advantage over Qwen2.5-coder for non-agentic tasks like just plain autocomplete and chat?
danielhanchen
Oh it should be better, especially since the model was specifically designed for coding tasks! You can disable the tool calling parts of the model!
andai
I've been reading about your dynamic quants, very cool. Does your library let me produce these, or only run them? I'm new to this stuff.
danielhanchen
Thank you! Oh currently not sadly - we might publish some stuff on it in the future!
Jayakumark
What will be the approx token/s prompt processing and generation speed with this setup on RTX 4090?
danielhanchen
I also just made IQ1_M which needs 160GB! If you have 160-24 = 136 ish of RAM as well, then you should get 3 tokens to 5 ish per second.
If you don't have enough RAM, then < 1 token / s
jdright
Any idea if there is a way to run on 256gb ram + 16gb vram with usable performance, even if barely?
danielhanchen
Yes! 3bit maybe 4bit can also fit! llama.cpp has MoE offloading so your GPU holds the active experts and non MoE layers, thus you only need 16GB to 24GB of VRAM! I wrote about how to do in this section: https://docs.unsloth.ai/basics/qwen3-coder#improving-generat...
jdright
awesome documentation, I'll try this. thank you!
pxc
> Qwen3-Coder is available in multiple sizes, but we’re excited to introduce its most powerful variant first
I'm most excited for the smaller sizes because I'm interested in locally-runnable models that can sometimes write passable code, and I think we're getting close. But since for the foreseeable future, I'll probably sometimes want to "call in" a bigger model that I can't realistically or affordably host on my own computer, I love having the option of high-quality open-weight models for this, and I also like the idea of "paying in" for the smaller open-weight models I play around with by renting access to their larger counterparts.
Congrats to the Qwen team on this release! I'm excited to try it out.
KronisLV
> I'm most excited for the smaller sizes because I'm interested in locally-runnable models that can sometimes write passable code, and I think we're getting close.
Likewise, I found that the regular Qwen3-30B-A3B worked pretty well on a pair of L4 GPUs (60 tokens/second, 48 GB of memory) which is good enough for on-prem use where cloud options aren't allowed, but I'd very much like a similar code specific model, because the tool calling in something like RooCode just didn't work with the regular model.
In those circumstances, it isn't really a comparison between cloud and on-prem, it's on-prem vs nothing.
callbacked
30B-A3B works extremely well as a generalist chat model when you pair with scaffolding such as web search. It's fast (for me) using my workstation at home running a 5070 + 128GB of DDR4 3200 RAM @ ~28 tok/s. Love MoE models.
Sadly it falls short during real world coding usage, but fingers crossed that a similarly sized coder variant of Qwen 3 can fill in that gap for me.
This is my script for the Q4_K_XL version from unsloth at 45k context:
llama-server.exe --host 0.0.0.0 --no-webui --alias "Qwen3-30B-A3B-Q4_K_XL" --model "F:\models\unsloth\Qwen3-30B-A3B-128K-GGUF\Qwen3-30B-A3B-128K-UD-Q4_K_XL.gguf" --ctx-size 45000 --n-gpu-layers 99 --slots --metrics --batch-size 2048 --ubatch-size 2048 --temp 0.6 --top-p 0.95 --min-p 0 --presence-penalty 1.5 --repeat-penalty 1.1 --jinja --reasoning-format deepseek --cache-type-k q8_0 --cache-type-v q8_0 --flash-attn --no-mmap --threads 8 --cache-reuse 256 --override-tensor "blk\.([0-9][02468])\.ffn_._exps\.=CPU"
pxc
I love Qwen3-30B-A3B for translation and fixing up transcripts generated by automatic speech recognition models. It's not the most stylish translator (a bit literal), but it's generally better than the automatic translation features built into most apps, and it's much faster since there's no network latency.
It has also been helpful (when run locally, of course) for addressing questions-- good faith questions, not censorship tests to which I already know the answers-- about Chinese history and culture that the DeepSeek app's censorship is a little too conservative for. This is a really fun use case actually, asking models from different parts of the world to summarize and describe historical events and comparing the quality of their answers, their biases, etc. Qwen3-30B-A3B is fast enough that this can be as fun as playing with the big, commercial, online models, even if its answers are not equally detailed or accurate.
jimmydoe
> good faith questions
yep, when you hire an immigrate software engineer, you don't ask them if Israel has a right to exist, or whether Vladivostok is part of china. Unless you are a DoD vendor which there won't be an interview anyway.
NitpickLawyer
Give devstral a try, fp8 should fit in 48GB, it was surprisingly good for a 24B local model, w/ cline/roo. Handles itself well, doesn't get stuck much, most of the things work OK (considering the size ofc)
KronisLV
I did! I do think Mistral models are pretty okay, but even the 4-bit quantized version runs at about 16 tokens/second, more or less usable but a biiiig step down from the MoE options.
Might have to swap out Ollama for vLLM though and see how different things are.
LinXitoW
Currently, the goal of everyone is creating one master model to rule them all, so we haven't seen too much specialization. I wonder how much more efficient smaller models could be if we created language specialized models.
It feels intuitively obvious (so maybe wrong?) that a 32B Java Coder would be far better at coding Java than a generalist 32B Coder.
californical
I’ll fill the role to push back on your Java coder idea!
First, Java code tends to be written a certain way, and for certain goals and business domains.
Let’s say 90% of modern Java is a mix of: * students learning to program and writing algorithms * corporate legacy software from non-tech focused companies
If you want to build something that is uncommon in that subset, it will likely struggle due to a lack of training data. And if you wanted to build something like a game, the majority of your training data is going to be based on ancient versions of Java, back when game development was more common in Java.
Comparatively, including C in your training data gives you exposure to a whole separate set of domain data for training, like IoT devices, kernels, etc.
Including Go will likely include a lot more networking and infrastructure code than Java would have had, which means there is also more context to pull from in what networking services expect.
Code for those domains follow different patterns, but the concepts can still be useful in writing Java code.
Now, there may be a middle ground where you could have a model that is still general for many coding languages, but given extra data and fine-tuning focused on domain-specific Java things — like more of a “32B CorporateJava Coder” model — based around the very specific architecture of Spring. And you’d be willing to accept that model to fail at writing games in Java.
It’s interesting to think about for sure - but I do feel like domain-specific might be more useful than language-specific
pxc
Don't we also find with natural languages that focusing on training data from only a single language doesn't actually result in better writing in the target language, either?
1899-12-30
jetbrains have done this with their mellum models that they use for autocompletion, https://ollama.com/JetBrains
fine tuned rather than created from scratch though.
larodi
Been using ggerganov’s llama vscode plugin with the smaller 2.5 models and it actually works super nice on a M3 Max
kimsia
I'm on a m1 max with 64gb ram, but i never use this vscode plugin before. Should I try?
Is this the one? https://github.com/ggml-org/llama.vscode it sems to be built for code completion rather than outright agent mode
larodi
It is RAG for your codebase, and provides code completion. The gain is the local inference, and is actually useful with smaller models.
The plugin itself provides chat also, but my gut feeling is that ggerganov runs several models at the some time, given he uses a 192gb machine.
Have not tried this scenario yet, but looking at my API bill I’m probably going to try 100% local dev at some point. Besides vibe coding with existing tools seems to not work that good for enterprise size codebases.
pxc
What languages do you work in? How much code do you keep? Do you end up using it as scaffolding and rewriting it, it leaving most of it as is?
larodi
Languages: JS/TS, C/C++, Shader Code, Some ESP Arduino code. Not counting all the boilerplate and CSS that I dont care about too much.
It very much reminds of tabbing autocomplete with IntelliSense step by step, but in a more diffusion-like way.
but my tool-set is a mixture of agentic and autocomplete, not 100% of each. I try to keep a clear focus of the architecture, and actually own the code by reading most of it, keeping straight the parts of the code the way i like.
segmondy
small models can never match bigger models, the bigger models just know more and are smarter. the smaller models can get smarter, but as they do, the bigger models get smart too. HN is weird because at one point this was the location where I found the most technically folks, and now for LLM I find them at reddit. tons of folks are running huge models, get to researching and you will find out you can realistically host your own.
pxc
> small models can never match bigger models, the bigger models just know more and are smarter.
They don't need to match bigger models, though. They just need to be good enough for a specific task!
This is more obvious when you look at the things language models are best at, like translation. You just don't need a super huge model for translation, and in fact you might sometimes prefer a smaller one because being able to do something in real-time, or being able to run on a mobile device, is more important than marginal accuracy gains for some applications.
I'll also say that due to the hallucination problem, beyond whatever knowledge is required for being more or less coherent and "knowing" what to write in web search queries, I'm not sure I find more "knowledgeable" LLMs very valuable. Even with proprietary SOTA models hosted on someone else's cloud hardware, I basically never want an LLM to answer "off the dome"; IME it's almost always wrong! (Maybe this is less true for others whose work focuses on the absolute most popular libraries and languages, idk.) And if an LLM I use is always going to be consulting documentation at runtime, maybe that knowledge difference isn't quite so vital— summarization is one of those things that seems much, much easier for language models than writing code or "reasoning".
All of that is to say:
Sure, bigger is better! But for some tasks, my needs are still below the ceiling of the capabilities of a smaller model, and that's where I'm focusing on local usage. For now that's mostly language-focused tasks entirely apart from coding (translation, transcription, TTS, maybe summarization). It may also include simple coding tasks today (e.g., fancy auto-complete, "ghost-text" style). I think it's reasonable to hope that it will eventually include more substantial programming tasks— even if larger models are still preferable for more sophisticated tasks (like "vibe coding", maybe).
If I end up having a lot of fun, in a year or two I'll probably try to put together a machine that can indeed run larger models. :)
saurik
> Even with proprietary SOTA models hosted on someone else's cloud hardware, I basically never want an LLM to answer "off the dome"; IME it's almost always wrong! (Maybe this is less true for others whose work focuses on the absolute most popular libraries and languages, idk.)
I feel like I'm the exact opposite here (despite heavily mistrusting these models in general): if I came to the model to ask it a question, and it decides to do a Google search, it pisses me off as I not only could do that, I did do that, and if that had worked out I wouldn't be bothering to ask the model.
FWIW, I do imagine we are doing very different things, though: most of the time, when I'm working with a model, I'm trying to do something so complex that I also asked my human friends and they didn't know the answer either, and my attempts to search for the answer are failing as I don't even know the terminology.
bredren
>you might sometimes prefer a smaller one because being able to do something in real-time, or being able to run on a mobile device, is more important than marginal accuracy gains for some applications.
This reminds me of ~”the best camera is the one you have with you” idea.
Though, large models are an http request away, there are plenty of reasons to want to run one locally. Not the least of which is getting useful results in the absence of internet.
larodi
All of these models are suitable for translation and that is what they are most suitable for. The architecture inherits from seq2seq and original transformers was created to benefit Google translations.
conradkay
For coding though it seems like people are willing to pay a lot more for a slightly better model.
mlyle
> HN is weird because at one point this was the location where I found the most technically folks, and now for LLM I find them at reddit.
Is this an effort to chastise the viewpoint advanced? Because his viewpoint makes sense to me: I can run biggish models on my 128GB Macbook but not huge ones-- even 2b quantized ones suck too many resources.
So I run a combination of local stuff and remote stuff depending upon various factors (cost, sensitivity of information, convenience/whether I'm at home, amount of battery left, etc ;)
Yes, bigger models are better, but often smaller is good enough.
Eggpants
The large models are using tools/functions to make them useful. Sooner or later open source will provide a good set of tools/functions for coding as well.
y1n0
I'd be interested in smaller models that were less general, with a training corpus more concentrated. A bash scripting model, or a clojure model, or a zig model, etc.
wkat4242
Well yes tons of people are running them but they're all pretty well off.
I don't have 10-20k$ to spend on this stuff. Which is about the minimum to run a 480B model, with huge quantisation. And pretty slow because for that price all you get is an old Xeon with a lot of memory or some old nvidia datacenter cards. If you want a good setup it will cost a lot more.
So small models it is. Sure, the bigger models are better but because the improvements come so fast it means I'm only 6 months to a year behind the big ones at any time. Is that worth 20k? For me no.
BriggyDwiggs42
The small model only needs to get as good as the big model is today, not as the big model is in the future.
ants_everywhere
There's a niche for small-and-cheap, especially if they're fast.
I was surprised in the AlphaEvolve paper how much they relied on the flash model because they were optimizing for speed of generating ideas.
ActorNightly
Not really true. Gemma from Google with quantized aware training does an amazing job.
Under the hood, the way it works, is that when you have final probabilities, it really doesn't matter if the most likely token is selected with 59% or 75% - in either case it gets selected. If the 59% case gets there with smaller amount of compute, and that holds across the board for the training set, the model will have similar performance.
In theory, it should be possible to narrow down models even smaller to match the performance of big models, because I really doubt that you do need transformers for every single forward pass. There are probably plenty of shortcuts you can take in terms of compute for sets of tokens in the context. For example, coding structure is much more deterministic than natural text, so you probably don't need as much compute to generate accurate code.
You do need a big model first to train a small model though.
As for running huge models locally, its not enough to run them, you need good throughput as well. If you spend $2k on a graphics card, that is way more expensive than realistic usage with a paid API, and slower output as well.
flakiness
The "qwen-code" app seems to be a gemini-cli fork.
https://github.com/QwenLM/qwen-code https://github.com/QwenLM/qwen-code/blob/main/LICENSE
I hope these OSS CC clones converge at some point.
Actually it is mentioned in the page:
we’re also open-sourcing a command-line tool for agentic coding: Qwen Code. Forked from Gemini Codemkagenius
Also, kudos to Gemini CLI team for making it open source (unlike claude) and that too easily tunable to new models like Qwen.
It would be great if it starts supporting other models too natively. Wouldn't require people to fork.
nicce
What seems to be typical these days is that big companies ship the first tool very fast, in poor condition (applies to Gemini CLI as well), and then let the OSS ecosystem fix the issues. Backend is closed so the app is their best shot. Then after some time the company gets the most credit and not all the contributors.
gavinray
I tried to use Jetbrains official Kotlin MCP SDK recently and it couldn't even serve the MCP endpoint on an URL that was different than what the default was expected to be...
They had made a bunch of hard-coded assumptions
mkagenius
> then let the OSS ecosystem fix the issues
That's precisely half of the point of OSS and I am pretty much okay with that.
rapind
I currently use claude-code as the director basically, but outsource heavy thinking to openai and gemini pro via zen mcp. I could instead use gemini-cli as it's also supported by zen. I would imagine it's trivial to add qwen-coder support if it's based on gemini-cli.
bredren
How was your experience using Gemini via Zen?
I’ve instead used a Gemini via plain ol’ chat, first building a competitive, larger context than Claude can hold then manually bringing detailed plans and patches to Gemini for feedback with excellent results.
I presumed mcp wouldn’t give me the focused results I get from completely controlling Gemini.
And that making CC interface via the MCP would also use up context on that side.
rapind
I just use it for architecture planning mostly when I want more info and to feed more info to claude. Tougher problems where 3 brains are better.
apwell23
what is the benefit of outsourcing to other models. do you see any noticable differences?
bredren
There are big gains to be had by having one top tier model review the work of another.
For example, you can drive one model to a very good point through several turns, and then have the second “red team” the result of the first.
Then return that to the first model with all of its built up context.
This is particularly useful in big plans doing work on complex systems.
Even with a detailed plan, it is not unusual for Claude code to get “stuck” which can look like trying the same thing repeatedly.
You can just stop that, ask CC to summarize the current problem and attempted solutions into a “detailed technical briefing.”
Have CC then list all related files to the problem including tests, then provide the briefing and all of the files to the second LLM.
This is particularly good for large contexts that might take multiple turns to get into Gemini.
You can have the consulted model wait to provide any feedback until you’ve said your done adding context.
And then boom, you get a detailed solution without even having to directly focus on whatever minor step CC is stuck on. You stay high level.
In general, CC is immediately cured and will finish its task. This is a great time to flip it into planning mode and get plan alignment.
Get Claude to output an update on its detailed plan including what has already been accomplished then again—-ship it to the consulting model.
If you did a detailed system specification in advance, (which CC hopefully was originally also working from) You can then ask the consulting model to review the work done and planned next steps.
Inevitably the consulting model will have suggestions to improve CC’s work so far and plans. Send it on back and you’re getting outstanding results.
ai-christianson
We shipped RA.Aid, an agentic evolution of what aider started, back in late '24, well before CC shipped.
Our main focuses were to be 1) CLI-first and 2) truly an open source community. We have 5 independent maintainers with full commit access --they aren't from the same org or entity (disclaimer: one has joined me at my startup Gobii where we're working on web browsing agents.)
I'd love someone to do a comparison with CC, but IME we hold our own against Cursor, Windsurf, and other agentic coding solutions.
But yes, there really needs to be a canonical FOSS solution that is not tied to any specific large company or model.
mrbonner
They also support Claude Code. But my understanding is Claude Code is closed source and only support Clade API endpoint. How do they make it work?
alwillis
But my understanding is Claude Code is closed source and only support Clade API endpoint. How do they make it work?
You set the environment variable ANTHROPIC_BASE_URL to an OpenAI-compatible endpoint and ANTHROPIC_AUTH_TOKEN to the API token for the service.
I used Kimi-K2 on Moonshot [1] with Claude Code with no issues.
There's also Claude Code Router and similar apps for routing CC to a bunch of different models [2].
mrbonner
That makes sense. Thanks Do you know if this works with AWS Berdrock as well? Or do I need to sort out to use the proxy approach?
Zacharias030
How good is it in comparison? This is an interesting apples to apples situation:)
vtail
Claude uses OpenAI-compatible APIs, and Claude Code respects environment variables that change the base url/token.
segmondy
no it doesn't, claude uses anthropic API. you need to run an anthropic2openAPI proxy
Imanari
You can use any model from openrouter with CC via https://github.com/musistudio/claude-code-router
chartered_stack
> I hope these OSS CC clones converge at some point.
Imo, the point of custom CLIs is that each model is trained to handle tool calls differently. In my experience, the tool call performance is wildly different (although they have started converging recently). Convergence is meaningful only when the models and their performance are commoditized and we haven't reached that stage yet.
danenania
I’ll throw out a mention for my project Plandex[1], which predates Claude Code and combines models from multiple providers (Anthropic, Google, and OpenAI by default). It can also use open source and local models.
It focuses especially on large context and longer tasks with many steps.
esafak
Have you measured and compared your agent's efficiency and success rate against anything? I am curious. It would help people decide; there are many coding agents now.
danenania
Working on it. I’m making a push currently on long horizon tasks, where Plandex already does well vs. alternatives, and plan to include side-by-side comparisons with the release.
carderne
Does Plandex have an equivalent to sub-agents/swarm or whatever you want to call it?
I’ve found getting CC to farm out to subagents to be the only way to keep context under control, but would love to bring in a different model as another subagent to review the work of the others.
danenania
It has built-in branches, which allow you to share context across as many related tasks as you want: https://docs.plandex.ai/core-concepts/branches
real-hacker
Yes. Just one open-source CC, with a configurable base_url/apikey, that would be great.
gabeyaw
Can you run qwen-code locally?
zkmon
At my work, here is a typical breakdown of time spent by work areas for a software engineer. Which of these areas can be sped up by using agentic coding?
05%: Making code changes
10%: Running build pipelines
20%: Learning about changed process and people via zoom calls, teams chat and emails
15%: Raising incident tickets for issues outside of my control
20%: Submitting forms, attending reviews and chasing approvals
20%: Reaching out to people for dependencies, following up
10%: Finding and reading up some obscure and conflicting internal wiki page, which is likely to be outdated
libraryofbabel
Really though? That’s only 2 hours per week writing code.
It’s true to say that time writing code is usually a minority of a developer’s work time, and so an AI that makes coding 20% faster may only translate to a modest dev productivity boost. But 5% time spent coding is a sign of serious organizational disfunction.
pyman
This is what software engineers need to be more productive:
- Agentic DevOps: provisions infra and solves platform issues as soon as a support ticket is created.
- Agentic Technical Writer: one GenAI agent writes the docs and keeps the wiki up to date, while another 100 agents review it all and flag hallucinations.
- Agentic Manager: attends meetings, parses emails and logs 24x7 and creates daily reports, shares these reports with other teams, and manages the calendar of the developers to shield them from distractions.
- Agentic Director: spots patterns in the data and approves things faster, without the fear of getting fired.
- Agentic CEO: helps with decision-making, gives motivational speeches, and aligns vision with strategy.
- Agentic Pet: a virtual mascot you have to feed four times a day, Monday to Friday, from your office's IP address. Miss a meal and it dies, and HR gets notified. (This was my boss's idea)
hexmiles
In case of holiday/sick leave do i need to find someone to feed the agentic pet from my ip address? Or is my manager responsability?
afiodorov
sign of serious organizational disfunction.
The reality is, most engineers spend far less than half their time writing new code. This is where the 80/20 principle comes into play. It's common for 80% of a company's revenue to come from 20% of its features. That core, revenue-generating code is often mature and requires more maintenance than new code. Its stability allows the company to afford what you call "dysfunction": having a large portion of engineers work on speculative features and "big bets" that might never see the light of day.
So, while it looks like a bug from a pure "coding hours" perspective, for many businesses, it's a strategic feature!
jameshart
I suspect a lot of that organizational dysfunction is related to a couple of things that might be changed by adjusting individual developer coding productivity:
1) aligning the work of multiple developers
2) ensuring that developer attention is focused only on the right problems
3) updating stakeholders on progress of code buildout
4) preventing too much code being produced because of the maintenance burden
If agentic tooling reduces the cost of code ownership, annd allows individual developers to make more changes across a broader scope of a codebase more quickly, all of this organizational overhead also needs to be revisited.
mpeg
IMHO, the biggest impact LLMs have had in my day to day has not been agentic coding. For example, meeting summarisers are great, it means I sometimes can skip a call or join while doing other things and I still get a list of bullet points afterwards.
I can point at a huge doc for some API and get the important things right away, or ask questions of it. I can get it to review PRs so I can quickly get the gist of the changes before digging into the code myself.
For coding, I don't find agents boost my productivity that much where I was already productive. However, they definitely allow me to do things I was unable to before (or would have taken very long as I wasn't an expert) – for example my type signatures have improved massively, in places where normally I would have been lazy and typed as any I now ask claude to come up with some proper types.
I've had it write code for things that I'm not great at, like geometry, or dataviz. But these are not necessarily increasing my productivity, they reduce my reliance on libraries and such, but they might actually make me less productive.
varispeed
Why would it be? I'd say it's the opposite. I someone keeps fiddling with the code majority of the time, it means they don't know what they are doing.
zelphirkalt
New requirements, new features, old bugs being fixed, refactoring code to improve maintainability, writing tests for edge cases previously not discovered, adapting code for different kinds of deployment, ...
Many reasons to touch existing code.
chrsw
I've been on embedded projects where several weeks of work were spent on changing one line of code. It's not necessarily organizational dysfunction. Sometimes it's getting the right data and the right deep understanding of a system, hardware/software interaction, etc, before you can make an informed change that affects thousands of people.
zkmon
Unfortunately it is true with any org that is rapidly reducing their risk appetite. It is not dysfunctional. It is about balancing the priorities at org level. Risk is distributed very thinly across many people. Heard of re-insurance business? sort of similar thing happens in software development as well.
zelphirkalt
It means though, that the business positions itself no longer as a software making business. No longer does it value being able to make software things that support its processes, whether those are customer processes or internal processes.
mathiaspoint
Serious organizational disfunction is a good way to describe most large tech companies.
rwmj
It doesn't if you have to manually check all that code. (Or even worse, you dump the code into a pull request and force someone else to manually check it - please do not do that.)
logsr
5% is pretty low but similar to what i have seen on low performing teams at 10K+ employee multinationals. this would also be why the vast majority of software today is bug ridden garbage that runs slower than the software we were using 20 years ago.
agentic coding will not fix these systemic issues caused by organizational dysfunction. agentic coding will allow the software created by these companies to be rewritten from scratch for 1/100th the cost with better reliability and performance though.
the resistance to AI adoption inside corporations that operate like this is intense and will probably intensify.
it takes a combination of external competitive pressure, investor pressure, attrition, PE takeovers, etc, to grind down internal resistance, which takes years or decades depending on the situation.
Too
> 1/100th the cost with better reliability and performance
Cheaper yes. More reliable? Absolutely not. Not with today’s models at least.
throwaw12
"10% running build pipelines + 20% submitting forms" vs 5% making code changes?
Are you in heavily regulated industry or dysfunctional organization?
Most big tech optimize their build pipelines a lot to reduce commit to deploy (or validation/test process) which keeps engineers focus on the same task while problem/solution is fresh.
khalic
How about you find out for yourself? Keep a chat window or an agent open and ask it how it could help with your tasks. My git messages and gitlab tickets are being written by AI for a year now, way better than anything I would half heartedly do on my side, really good commit messages too. Claude even reminds me to create/update the ticket.
paffdragon
I find the commits written by AI often inadequate, as they mostly just describe what is already in the diff, but miss the background on why was the change needed, why this approach was chosen, etc, the important stuff...
khalic
Then ask it to write the commit differently, or you can explain why in the prompt. Edit: I start by creating the ticket with Claude+terminal tool, the title and descriptions gives context info to the llm, then we do the task, then commit and update the ticket
carderne
Do you feed the LLM additional context for the commit message, or it is just summarising what’s in the commit? In the latter case, what’s the point? The reader can just get _their_ LLM to do a better job.
In the former case… I’m interested to hear how they’re better? Do you choose an agent with the full context of the changes to write the message, so it knows where you started, why certain things didn’t work? Or are you prompting a fresh context with your summary and asking it to make it into a commit message? Or something else?
khalic
Depends, I have a prompt ready for changes I made manually, that checks the diff, gets the context, spits a conventional commit with a summary of the changes, I check, correct if needed and add the ticket number. It’s faster because it types really fast, no time thinking about phrasing and remembering the changes, and usually way more complete then what I would have written, given time constraints.
If I’m using a CLI:
the agent already has: - the context from the chat - the ticket number via me or when it created the ticket - meta info via project memory or other terminal commands like API call etc - Info on commit format from project memory
So it boils down to asking it to commit and update the ticket when we’re done with the task in that case. Having a good workflow is key
For your question: I still read and validated/correct, in the end I’m the one committing the code! So it’s the usual requirements from there. If someone would use their LLM the results would vary, here they have an approved summary. This is why human in the loop is essential.
hansmayer
[flagged]
Eisenstein
You are saying that people need to write so complex that an LLM that can pass an LSAT test with flying colors is unable to summarize its changes in a few sentences, or else their work is not critical? That is a high bar.
khalic
[flagged]
rwmj
We must have the same job! Generating code is a miniscule part of my job. We have the same level of organizational disfunction. Mostly the work part involves long investigations of customer bugs and long face to face calls with customers - I'm only getting the stuff that stumped level 1 and level 2 support.
I actually tried to use Qwen3[1] to analyse customer cases and it was worse than useless at it.
[1] We can't use any online model as these bug reports contain large amounts of PII, customer data, etc.
mhl47
In theory, nearly all of them?
Many of those things could be improved today without AI but e.g. raising Incidents for issues outside of your control could also give you a suggestion already that you just have to tick off.
Not saying we are there yet but hard to imagine it's not possible.
zkmon
Raising incidents is not about suggestions. Things like build pipelines run into issues, someone from Ops need to investigate, and maybe bump up some pods or apply some config changes on their end. Or some wiki page has conflicting information, someone need to update it with correct information after checking with the relevant other people, policies and standards. The other people might be on vacation and their delegate misguides as they are not aware of the recently changed process.
It's probably messier than you think.
JimmaDaRustla
Your place is work sucks
Also, you're not making an argument against agentic coding, you're actually making an argument for it - you don't have time to code, so you need someone or something to code for you.
sannysanoff
You should automate this, like i did. You're an engineer, no? Work around the digital bureaucracy.
- Running build pipelines: make cli tool to initiate them, monitor them and notify you on completion/error (audio). Allows to chain multiple things. Run in background terminal.
- Learning about changed process and people via zoom calls, teams chat and emails: pass logs of chats and emails to LLM with particular focus. Demand zoom calls transcripts published for that purposes (we use meet)
- Raising incident tickets for issues outside of my control: automate this with agent: allow it to access as much as needed, and guide it with short guidance - all doable via claude code + custom MCP
- Submitting forms, attending reviews and chasing approvals - best thing to automate. They want forms? They will have forms. Chasing approvals - fire and forget + queue management, same.
- Reaching out to people for dependencies, following up: LLM as personal assistant is classic job. Code this away.
- Finding and reading up some obscure and conflicting internal wiki page, which is likely to be outdated: index all data and put it into RAG, let agent dig deeper.
Most of the time you spend is on scheduling micro-tasks, switching between them and maintaining unspoken queue of checking various saas frontends. Formalize micro-task management, automate endpoints, and delegate it to your own selfware (ad-hoc tools chain you vibe coded for yourself only, tailored for particular working environment).
I do this all (almost) to automate away non-coding tasks. Life is fun again.
Hope this helps.
nisten
I've been using it all day, it rips. Had to bump up toolcalling limit in cline to 100 and it just went through the app no issues, got the mobile app built, fixed throug hthe linter errors... wasn't even hosting it with the toolcall template on with the vllm nightly, just stock vllm it understood the toolcall instructions just fine
MaxikCZ
Im interested in more info? Where do you host it? Whats the hardware, and exact model? What t/s do you get? What is the codebase size? Etc pls, thank you
ramoz
Nice, what model & on what hardware?
nxobject
Welp, time to switch aider models for the _second_ time in a week...
manmal
How good is it at editing files? Many write/replace errors?
apwell23
so are you tell where you hosted it?
nnx
This suggests adding a `QWEN.md` in the repo for agents instructions. Where are we with `AGENTS.md`? In a team repo it's getting ridiculous to have a duplicate markdown file for every agent out there.
singhrac
I just symlink to AGENTS.md, the instructions are all the same (and gitignore the model-specific version).
sunaookami
I just make a file ".llmrules" and symlink these files to it. It clutters the repo root, yes...
oblio
Can't these hyper-advanced-super-duper tools discover what UNIX tools since circa 1970 knew, and just have a flag/configuration setting pointing them to the config file location? Excuse me if they already do :-)
In which case you'd have 1 markdown file and at least for the ones that are invoked via the CLI, just set up a Makefile entry point that leads them to the correct location.
drewbitt
CLAUDE.md MISTRAL.md GEMINI.md QWEN.md GROK.md .cursorrules .windsurfrules .copilot-instructions
Saw a repo recently with probably 80% of those
falcor84
It would be funny to write conflicting instructions on these, and then unleash different coding agents on the same repo in parallel, and see which one of them first identifies the interference from the others and rewrites their instructions to align with its own.
yard2010
Lol you can even tell each model to maliciously and secretly sabotage other agents and see which one wins.
czottmann
I built https://github.com/czottmann/render-claude-context for that exact reason.
> This node.js CLI tool processes CLAUDE.md files with hierarchical collection and recursive @-import resolution. Walks directory tree from current to ~/.claude/, collecting all CLAUDE.md files and processing them with file import resolution. Saves processed context files with resolved imports next to the original CLAUDE.md files or in a specific location (configurable).
I mostly use Claude Code, but every now and then go with Gemini, and having to maintain two sets of (hierarchical) instructions was annoying. And then opencode showed up, which meant yet another tool I wanted to try out and …well.
mattigames
Maybe there could be an agent that is in charge of this and it's trained to automatically create a file for any new agent, it could even temporarily delete local copies of MD files that no agents are using at the moment to free the visual clutter when navigating the repo.
theshrike79
I tried making an MCP with the common shit I need to tell the agents, but it didn't pan out.
Now I have a git repo I add as a submodule and tell each tool to read through and create their own WHATEVER.md
redhale
https://github.com/intellectronica/ruler
Library to help with this. Not great that a library is necessary, but useful until this converges to a standard (if it ever does).
apwell23
these files are for free publicity on github
indigodaddy
How does one keep up with all this change? I wish we could fast-forward like 2-3 years to see if an actual winner has landed by then. I feel like at that point there will be THE tool, with no one thinking twice about using anything else.
segmondy
One keeps up with it, by keeping up with it. Folks keep up with latest social media gossip, the news, TV shows, or whatever interests them. You just stay on it. Weekend I got to running Kimi K2, last 2 days I have been driving Ernie4.5-300B, Just finished downloading the latest Qwen3-235b this morning and started using it this evening. Tonight I'll start downloading this 480B, might take 2-3 days with my crappy internet and then I'll get to it.
Obsession?
Sabinus
Do you write about your assessments of model capabilities and the results of your experiments?
Zacharias030
what kind of hardware do you run it on?
SchemaLoad
Just ignore it until something looks useful. There's no reason to keep up, it's not like it takes 3 years experience to type in a prompt box.
yard2010
> it's not like it takes 3 years experience to type in a prompt box
This should be written on the coffin of full stack development.
barrell
Yeah second this. I find model updates mildly interesting, but besides grok 4 I haven’t even tried a new model all year.
Its a bit like the media cycle. The more jacked in you are, the more behind you feel. I’m less certain there will be winners as much as losers, but for sure the time investment on staying up to date on these things will not pay dividends to the average hn reader
stets
I'm using claude code and making stuff. I'm keeping an eye and being aware of these new tools but I wait for the dust to settle and see if people switch or are still hyped after the hype dies down. X / HackerNews are good for keeping plugged in.
blibble
don't bother at all
assuming it doesn't all implode due to a lack of profitability, it should be obvious
aitchnyu
The underlying models are apparently profitable. Inference costs are in a exponential fall that makes Gordon Moore faint. OpenRouter shows Anthropic, AWS, Google host Claude at same rates, apparently nobody is price dumping.
That said, code+git+agent is only acceptable way for technical staff to interact with AI. Tools with sparkles button can go to hell.
https://a16z.com/llmflation-llm-inference-cost/ https://openrouter.ai/anthropic/claude-sonnet-4
blibble
if I dropped 99.999999% of my costs I'd be Google level profitable too
oblio
I was thinking this exact same thing last night.
We don't actually need a winner, we need 2-3-4 big, mature commercial contenders for the state of the art stuff, and 2-3-4 big, mature Open Source/open weights models that can be run on decent consumer hardware at near real-time speeds, and we're all set.
Sure, there will probably be a long tail, but the average programmer probably won't care much about those, just like they don't care about Erlang, D, MoonScript, etc.
lsllc
I think in 2-3 years, it'll be the same story except it'll be bigger/better/faster.
As Heraclitus said "The only constant in life is change"
(and maybe Emacs)
theptip
Things will be moving faster in 2-3 years most likely. (The recursive self-improvement flywheel is only just starting to pick up momentum, and we’ll have much more LLM inference compute available.)
Figuring out how to stay sane while staying abreast of developments will be a key skill to cultivate.
I’m pretty skeptical there will be a single model with a defensible moat TBH. Like cloud compute, there is both economy of scale and room for multiple vendors (not least because bigco’s want multiple competing bids).
oblio
I'm actually waiting for something different - a "good enough" level for programming LLMs:
1. Where they can be used as autocompletion in an IDE at speeds comparable with Intellisense 2. And where they're good enough to generate most code reliably, while using a local LLM 3. While running on hardware costing in total max 2000€ 4. And definitely with just a few "standard" pre-configured Open Source/open weights LLMs where I don't have to become an LLM engineer to figure out the million knobs
I have no clue how Intellisense works behind the scenes, yet I use it every day. Same story here.
theptip
“Good enough” will be like programming languages; an evolving frontier with many choices. New developments will make your previous “good enough” look inadequate.
Given how much better the bleeding edge models are now than 6 months ago, as long as any model is getting smarter I don’t see stagnation as a possibility. If Gemini starts being better at coding than Claude, you’re gonna switch over if your livelihood is dependent on it.
3abiton
It depends on the level of 'keeping up'. I follow the news, but it's impossible to dip your toe in every new model. Some stick around, but the majority pass through.
chisleu
I tried using the "fp8" model through hyperbolic but I question if it was even that model. It was basically useless through hyperbolic.
I downloaded the 4bit quant to my mac studio 512GB. 7-8 minutes until first tokens with a big Cline prompt for it to chew on. Performance is exceptional. It nailed all the tool calls, loaded my memory bank, and reasoned about a golang code base well enough to write a blog post on the topic: https://convergence.ninja/post/blogs/000016-ForeverFantasyFr...
Writing blog posts is one of the tests I use for these models. It is a very involved process including a Q&A phase, drafting phase, approval, and deployment. The filenames follow a certain pattern. The file has to be uploaded to s3 in a certain location to trigger the deployment. It's a complex custom task that I automated.
Even the 4bit model was capable of this, but was incapable of actually working on my code, prefering to halucinate methods that would be convenient rather than admitting it didn't know what it was doing. This is the 4 bit "lobotomized" model though. I'm excited to see how it performs at full power.
jasonthorsness
What sort of hardware will run Qwen3-Coder-480B-A35B-Instruct?
With the performance apparently comparable to Sonnet some of the heavy Claude Code users could be interested in running it locally. They have instructions for configuring it for use by Claude Code. Huge bills for usage are regularly shared on X, so maybe it could even be economical (like for a team of 6 or something sharing a local instance).
danielhanchen
I'm currently trying to make dynamic GGUF quants for them! It should use 24GB of VRAM + 128GB of RAM for dynamic 2bit or so - they should be up in an hour or so: https://huggingface.co/unsloth/Qwen3-Coder-480B-A35B-Instruc.... On running them locally, I do have docs as well: https://docs.unsloth.ai/basics/qwen3-coder
zettabomb
Any significant benefits at 3 or 4 bit? I have access to twice that much VRAM and system RAM but of course that could potentially be better used for KV cache.
danielhanchen
So dynamic quants like what I upload are not actually 4bit! It's a mixture of 4bit to 8bit with important layers being in higher precision! I wrote about our method here: https://docs.unsloth.ai/basics/unsloth-dynamic-2.0-ggufs
sourcecodeplz
For coding you want more precision so the higher the quant the better. But there is discussion if a smaller model in higher quant is better than a larger one in lower quant. Need to test for yourself with your use cases I'm afraid.
e: They did announce smaller variants will be released.
jychang
You definitely want to use 4bit quants at minimum.
https://arxiv.org/abs/2505.24832
LLMs usually have about 3.6 bits of data per parameter. You're losing a lot of information if quantized to 2 bits. 4 bit quants are the sweet spot where there's not much quality loss.
fzzzy
I would say that three or four bit are likely to be significantly better. But that’s just from my previous experience with quants. Personally, I try not to use anything smaller than a Q4.
simonw
There's a 4bit version here that uses around 272GB of RAM on a 512GB M3 Mac Studio: https://huggingface.co/mlx-community/Qwen3-Coder-480B-A35B-I... - see video: https://x.com/awnihannun/status/1947771502058672219
That machine will set you back around $10,000.
jychang
You can get similar performance on an Azure HX vm:
https://learn.microsoft.com/en-us/azure/virtual-machines/siz...
osti
How? These don't even have GPU's right?
kentonv
Ugh, why is Apple the only one shipping consumer GPUs with tons of RAM?
I would totally buy a device like this for $10k if it were designed to run Linux.
jauntywundrkind
Intel already has a great value GPU. Everyone wants them to disrupt the game, destroy the product niches. It's general purpose compute performance is quite ass alas but maybe that doesn't matter for AI?
I'm not sure if there are higher capacity gddr6 & 7's rams to buy. I semi doubt you can add more without more channels, to some degree, but also, AMD just shipped R9700 based on rx9070 but with double the ram. But something like Strix Halo, an API with more lpddr channels could work. Word is that Strix Halo's 2027 successor Medusa Halo will go to 6 channels and it's hard to see a significant advantage without that win; the processing is already throughput constrained-ish and a leap on memory bandwidth will definitely be required. Dual channel 128b isn't enough!
There's also MRDIMMs standard, which multiplexes multiple chips. That promises a doubling of both capacity and throughout.
Apple's definitely done two brilliant costly things, by putting very wide (but not really fast) memory on package (Intel had dabbled in doing similar with regular width ram in consumer space a while ago with Lakefield). And then by tiling multiple cores together, making it so that if they had four perfect chips next to each other they could ship it as one. Incredibly brilliant maneuver to get fantastic yields, and to scale very big.
sbrother
You can buy a RTX 6000 Pro Blackwell for $8000-ish which has 96GB VRAM and is much faster than the Apple integrated GPU.
sagarm
You can get 128GB @ ~500GB/s now for ~$2k: https://a.co/d/bjoreRm
It has 8 channels of DDR5-8000.
gaspoweredcat
i mean sure its not quite 512gb levels but you can get 128gb on a ryzen AI max chipset which has unified memory like apple, theyre also pretty reasonably priced, i saw an AI max 370 with 96gb on amazon earlier for a shade over £1000, guess you could boost that with an eGPU to gain a bit extra but 64gb would likely be the max you could add so still not quite enough to run full qwen3 coder at a decent quant but not far off, hopefully the next gen will offer more ram or another model comes out that can beat Q3 with fewer params
ashvardanian
That's very informative, thanks! So a DGX H200 should be able to run it at 16-bit precision. If I recall correctly, the current hourly rate should be around $25. Not sure what the throughput is, though.
ilaksh
To run the real version with the bench arks they give, it would be a nonquantized non distilled version. So I am guessing that is a cluster of 8 H200s if you want to be more or less up to date. They have B200s now which are much faster but also much more expensive. $300,000+
You will see people making quantized distilled versions but they never give benchmark results.
danielhanchen
Oh you can run the Q8_0 / Q8_K_XL which is nearly equivalent to FP8 (maybe off by 0.01% or less) -> you will need 500GB of VRAM + RAM + Disk space. Via MoE layer offloading, it should function ok
summarity
This should work well for MLX Distributed. The low activation MoE is great for multi node inference.
ilaksh
1. What hardware for that. 2. Can you do a benchmark?
sourcecodeplz
With RAM you would need at least 500gb to load it but some 100-200gb more for context too. Pair it with a 24gb GPU and the speed will be 10t/s, at least, I estimate.
danielhanchen
Oh yes for the FP8, you will need 500GB ish. 4bit around 250GB - offloading MoE experts / layers to RAM will definitely help - as you mentioned a 24GB card should be enough!
vFunct
Do we know if the full model is FP8 or FP16/BF16? The hugging face page says BF16: https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct
So likely it needs 2x the memory.
chisleu
A Mac Studio 512GB can run it in 4bit quantization. I'm excited to see unsloth dynamic quants for this today.
827a
The initial set of prices on OpenRouter look pretty similar to Claude Sonnet 4, sadly.
btian
Do need to be super fancy. Just RTX Pro 6000 and 256GB of RAM.
chisleu
A mac studio can run it at 4bit. Maybe at 6 bit.
rbren
Glad to see everyone centering on using OpenHands [1] as the scaffold! Nothing more frustrating than seeing "private scaffold" on a public benchmark report.
swyx
more info on AllHands from robert (above) https://youtu.be/o_hhkJtlbSs
KaoruAoiShiho
How is cognition so incompetent? They got hundreds of millions of dollars but now they're not just supplanted by Cursor and Claude Code but also by their literal clone, an outfit that was originally called "OpenDevin".
samrus
The AI space is attracting alot of grifters. Even the initial announcement for devin was reaking of elon musk style overpromising.
Im sure the engineers are doing the best work they can. I just dont think leadership is as interested in making a good product as they are in creating a nice exit down the line
incomingpain
I just finally got devstral working well.
Openhands is clearly the best ive used so far. Even gemini cli is lesser.
ramon156
Are you purposefully ignoring Zed?
rapind
I just checked and it's up on OpenRouter. (not affiliated) https://openrouter.ai/qwen/qwen3-coder
generalizations
> Additionally, we are actively exploring whether the Coding Agent can achieve self-improvement
How casually we enter the sci-fi era.
yakz
I don’t get the feeling that the amount of money being spent is at all casual.
jasonjmcghee
We have self driving cars, humanoid robots, and thinking machines. I think we're there.
1dom
Casual and safe daily use of hoverboards and meal-in-a-pill are my indicators. I think we're not quite there yet, but everyone's different!
cvs268
...and "No roads". Don't forget no roads! :-)
undefined
undefined
Get the top HN stories in your inbox every day.
I'm currently making 2bit to 8bit GGUFs for local deployment! Will be up in an hour or so at https://huggingface.co/unsloth/Qwen3-Coder-480B-A35B-Instruc...
Also docs on running it in a 24GB GPU + 128 to 256GB of RAM here: https://docs.unsloth.ai/basics/qwen3-coder