Rendered at 07:24:10 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
SeriousM 12 hours ago [-]
> Anthropic wants firms like Alibaba punished
Do they care what we want? Im happy that China is copying the models and make them available for everyone, with 1/10 of it's price.
bicepjai 11 hours ago [-]
Do they hear themselves ? We steal better, but don’t steal from us :)
forestry 7 hours ago [-]
Whatever you thoughts on the ethics of the original training, are you comfortable with innovation drying up? Can’t make new models if revenue isn’t coming in.
khurs 6 hours ago [-]
Why would innovation dry up when there is more competition?
a_sewer_rat 11 hours ago [-]
+1.
Ex-product manager in the SF Bay. Retooling as a product engineer/ product builder because I can build and tinker confidently with LLM assistance, and I want to reduce my dependencies.
So I’m:
* prioritizing local Qwen codegen so I can cancel my OpenAI subscription.
* watching CS lectures and learning the fundamentals from the ground up.
Maybe it’s the long-term unemployment talking, but I think the open-model strategy is winning and will continue to win. The recent stories about AI budgets are just the beginning. Wait until executives realize that the advertised “$X.XX per million tokens” is only part of the bill. Context windows, system prompts, conversation history, RAG, tool calls, and orchestration all add up.
Cost-conscious companies will have a strong incentive to hire engineers who can run, customize, and maintain open models rather than paying indefinitely for closed APIs.
bofadeez 12 hours ago [-]
Competition is supposed to be the safety net of capitalism that ensures high quality and low prices. Distillation is not illegal. They're just using the LLM in a lawful way. So yes we all should thank the capitalist Chinese labs for helping the consumer in a win-win fashion.
This military strategy talk is not interesting to anyone other than Anthropic AI researchers who decide to work there.
recursivegirth 12 hours ago [-]
This. I find distillation way more ethical than piracy. The U.S. already jumped the gun by making this stuff not copyrightable. We should of instead held these companies liable. It's not like the research they did in the process was for naught. They (U.S. AI companies) could of sourced legal material without stooping to the same lows that China and other nations are willing to go to. All in the name of a buck.
Anyways - I pay for z.ai, Opencode Go, Ollama Cloud. I love me some distilled models.
nerdsniper 6 hours ago [-]
* should "have"
* could "have"
recursivegirth 6 hours ago [-]
Sorry, should have ran my comment through a LLM to appease the irreverent assholes, forgive me sire.
bdavid21wnec 13 hours ago [-]
I don’t know much about the ins and outs of LLMs, but if something can be reversed engineered that quickly. What exactly is the large moat that makes them worth 1trillion?
Daishiman 12 hours ago [-]
They have none. Google came to the same conclusion years ago.
d0mine 14 hours ago [-]
How many accounts Anthropic used to mine the web
androiddrew 14 hours ago [-]
Yeah. The "I don't care" line from The Fugitive comes to mind.
andrewinardeer 10 hours ago [-]
True, but I'm not jumping from the outlet pipe.
teddyX 14 hours ago [-]
Tartigrade_playing_a_violin.gif
pixiemaster 13 hours ago [-]
you can reverse engineer the whole thing with just 25k accounts?
don’t let the stock market hear that.
gentlewater 12 hours ago [-]
The stock market rewarded SpaceX with a record breaking IPO valuation for promises that it would too offer a thing that could be reverse engineered with just 25k accounts. Possibly in space. The stock market doesn’t care.
adastra22 12 hours ago [-]
SpaceX’s value proposition is far more than AI.
victorbjorklund 6 hours ago [-]
They do claim most of their business is AI though with some rockets sprinkled on top
adastra22 5 hours ago [-]
Most of their forecasted revenue within the time horizon wall st looks at.
SpaceX is making a gamble that will lead to quadrillion dollar valuations if things work out.
Yes, I recognize that is larger than the entire economy of the earth. That’s precisely the point.
moogly 11 hours ago [-]
Far more ludicrous, that is.
SonOfKyuss 15 hours ago [-]
Actors, musicians, and writers are all playing the world’s smallest violin right now.
happyPersonR 14 hours ago [-]
This. Lolllll
It’s so sad no one is going to shed a single tear
wookmaster 7 hours ago [-]
Ok? Claude is already mining everything else. Why should they be protected but not the data they stole ?
gruturo 12 hours ago [-]
The "Oh no! Anyway" meme from Top Gear has never been more appropriate.
Oh no! The company that both stole all it's training data and claims to want to build the most ethically oriented machine cares whether other humans are using it to do their own thing!
It kinda seems like Anthropic and the trump administration are doing some bizarre kabuki theatre.
hogwasher 12 hours ago [-]
Neither Anthropic nor anyone in the Trump administration can hold a candle to kabuki theatre.
(I know that's not what you meant, but still! Art, the LLM and the Trump admin cannot.)
Anyway yeah this is beyond absurd of them to fuss about this. Oh no, the thieves that stole from checks notes basically everbody have been robbed of the stolen goods and nooow they care about copyright, huh, how shocking. I'm sure they'll settle all their court cases with artists and admit fault now, right? ...right?
It is too late, the Qwen 3.7 Max is above both Gemini 3.1 and Opus 4.8. And Qwen 3.7 Plus is on level with Fable. There is literally no point of paying exorbitant API prices to US Labs.
And remember: it's not only about coding. Those "selected US companies" to gain access to Mythos now are not using it to replace coders. And Qwen is a solid answer. (The next is GLM 5.2, but not there yet - Qwen 3.7 made a significant leap forward.)
Do they care what we want? Im happy that China is copying the models and make them available for everyone, with 1/10 of it's price.
Ex-product manager in the SF Bay. Retooling as a product engineer/ product builder because I can build and tinker confidently with LLM assistance, and I want to reduce my dependencies.
So I’m:
* prioritizing local Qwen codegen so I can cancel my OpenAI subscription. * watching CS lectures and learning the fundamentals from the ground up.
Maybe it’s the long-term unemployment talking, but I think the open-model strategy is winning and will continue to win. The recent stories about AI budgets are just the beginning. Wait until executives realize that the advertised “$X.XX per million tokens” is only part of the bill. Context windows, system prompts, conversation history, RAG, tool calls, and orchestration all add up.
Cost-conscious companies will have a strong incentive to hire engineers who can run, customize, and maintain open models rather than paying indefinitely for closed APIs.
This military strategy talk is not interesting to anyone other than Anthropic AI researchers who decide to work there.
Anyways - I pay for z.ai, Opencode Go, Ollama Cloud. I love me some distilled models.
* could "have"
don’t let the stock market hear that.
SpaceX is making a gamble that will lead to quadrillion dollar valuations if things work out.
Yes, I recognize that is larger than the entire economy of the earth. That’s precisely the point.
It’s so sad no one is going to shed a single tear
https://knowyourmeme.com/memes/oh-no-anyway
It kinda seems like Anthropic and the trump administration are doing some bizarre kabuki theatre.
(I know that's not what you meant, but still! Art, the LLM and the Trump admin cannot.)
Anyway yeah this is beyond absurd of them to fuss about this. Oh no, the thieves that stole from checks notes basically everbody have been robbed of the stolen goods and nooow they care about copyright, huh, how shocking. I'm sure they'll settle all their court cases with artists and admit fault now, right? ...right?
Yeah no, fuck 'em.
And remember: it's not only about coding. Those "selected US companies" to gain access to Mythos now are not using it to replace coders. And Qwen is a solid answer. (The next is GLM 5.2, but not there yet - Qwen 3.7 made a significant leap forward.)