Summarizer

Open Source Alternative Interest

Comments about preferring local inference, interest in Chinese open-source models, suggestions that proprietary restrictions push users toward open weights solutions

← Back to Claude Opus 4.7

Driven by a growing frustration with the restrictive "safety" filtering and perceived arrogance of Silicon Valley’s proprietary giants, many users are gravitating toward high-performance open-weights models like Qwen that can be run on local hardware. These enthusiasts argue that local inference provides essential control over privacy, parameters, and predictability, often matching or exceeding the capabilities of closed models while avoiding the "enshittification" of the open internet. Geopolitical concerns and ethical pushback against U.S.-centric corporate policies are further fueling interest in Chinese and European alternatives as a necessary hedge against being at the mercy of external providers. Ultimately, the community views the shift toward open-source solutions not just as a technical preference, but as a vital survival strategy to ensure that powerful AI capabilities remain accessible and uncensored.

22 comments tagged with this topic

View on HN · Topics
What about Qwen? Does it get that right?
View on HN · Topics
I've run several local models that get this right. Qwen 3.5 122B-A10B gets this right, as does Gemma 4 31B. These are local models I'm running on my laptop GPU (Strix Halo, 128 GiB of unified RAM). And I've been using this commonly as a test when changing various parameters, so I've run it several times, these models get it consistently right. Amazing that Opus 4.6 whiffs it.
View on HN · Topics
If you ask claude in chinese it thinks its deepseek.
View on HN · Topics
Yep, that is exactly what happens. It's a disgrace that their models aren't open, after training on everything humanity has preserved. They should at least release the weights of their old/deprecated models, but no, that would be losing money.
View on HN · Topics
Most likely, would be cool yes see a open source Nivel use diffusion for thinking.
View on HN · Topics
... here's the pelican, I think Qwen3.6-35B-A3B running locally did a better job! https://simonwillison.net/2026/Apr/16/qwen-beats-opus/
View on HN · Topics
You can run GPT2! Here's the medium model: https://huggingface.co/openai-community/gpt2-medium I will now have it continue this comment: I've been running gps for a long time, and I always liked that there was something in my pocket (and not just me). One day when driving to work on the highway with no GPS app installed, I noticed one of the drivers had gone out after 5 hours without looking. He never came back! What's up with this? So i thought it would be cool if a community can create an open source GPT2 application which will allow you not only to get around using your smartphone but also track how long you've been driving and use that data in the future for improving yourself...and I think everyone is pretty interested. [Updated on July 20] I'll have this running from here, along with a few other features such as: - an update of my Google Maps app to take advantage it's GPS capabilities (it does not yet support driving directions) - GPT2 integration into your favorite web browser so you can access data straight from the dashboard without leaving any site! Here is what I got working. [Updated on July 20]
View on HN · Topics
I agree with what you what you have written, which is why I would never pay a subscription to an external AI provider. I prefer to run inference on my own HW, with a harness that I control, so I can choose myself what compromise between speed and the quality of the results is appropriate for my needs. When I have complete control, resulting in predictable performance, I can work more efficiently, even with slower HW and with somewhat inferior models, than when I am at the mercy of an external provider.
View on HN · Topics
Thing is that Anthropic was always working with DoD, too, and the line in the sand they drew looked really noble until I found it didn't not apply to me, a non-US citizen. Dario made it clear that was the case. And so the difference, to me, was irrelevant. I'll buy based on value, and keep a poker in the fire of Chinese & European open weight models, as well.
View on HN · Topics
Dario in fact said it was ok to spy and drone non-US citizens, and in fact endorsed American foreign policy generally. So, no, I'm not voting with my wallet for one American country versus the other. I'll pick the best compromise product for me, and then also boost non-American R&D where I can.
View on HN · Topics
There is zero cost to switching ai models. Paid or open source. It's one line mostly.
View on HN · Topics
This is one of the many reasons I don't think the model companies are going to win the application space in coding. There's literally zero context lost for me in switching between model providers as a cursor user at work. For personal stuff I'll use an open source harness for the same reason.
View on HN · Topics
I think this is more about which model you steer your coding harness to. You can also self-host a UI in front of multiple models, then you own the chat history.
View on HN · Topics
I predict this sort of filtering is only going to get worse. This will probably be remembered as the 'open internet' era of LLMs before everything is tightly controlled for 'safety' and regulations. Forcing software devs to use open source or local models to do anything fun.
View on HN · Topics
> Forcing software devs to use open source or local models to do anything fun. Episode Five-Hundred-Bazillenty-Eight of Hacker News: the gang learns a valuable lesson after getting arrested at an unchaperoned Enshittification party and having to call Open Source to bail them out.
View on HN · Topics
May not be very effective if so. I'm assuming finding vulnerabilities in open source projects is the hard part and what you need the frontier models for. Writing an exploit given a vulnerability can probably be delegated to less scrupulous models.
View on HN · Topics
Codex or the Chinese models
View on HN · Topics
It appears we're learning the hard way that we can't rely on capabilities of models that aren't open weights. These can be taken from us at any time, so expect it to get much worse..
View on HN · Topics
Can't wait for a random chinese company to train a model on Mythos by breaking Anthropic's ToS just to release it for free and with open weights.
View on HN · Topics
Qwen 3.6 OSS and now this, almost feels like Anthropic rushed a release to steal hype away from Qwen
View on HN · Topics
> First, Opus 4.7 uses an updated tokenizer that improves how the model processes text wow can I see it and run it locally please? Making API calls to check token counts is retarded.
View on HN · Topics
> indeed, during its training we experimented with efforts to differentially reduce these capabilities can't wait for the chinese models to make arrogant silicon valley irrelevant