> we are open-sourcing two foundation models co-developed by Hy4 and Hyra 2.0: HyCell for single-cell representation and HyPDE for physical field modeling.
What a refreshing contrast. Where American startups brag about stopping researchers from writing grants about neglected tropical diseases and making novel anti-depressants, these folks are open sourcing the means to do science. https://news.ycombinator.com/item?id=49651727
That's how you get a civilizational advantage. By allowing people to just do things. It's so bizarre to watch China embrace the wonders of science and technology, while certain American startups and communities repeat "WHAT ABOUT THE BIOWEAPONS?"
> It's so bizarre to watch China embrace the wonders of science and technology, while certain American startups and communities repeat "WHAT ABOUT THE BIOWEAPONS?"
Eh, I'm reasonably certain it's not altruism. It's another form of competition with the US companies and I'm thankful for it while it lasts.
With China's Announcement 48 (export restrictions on AI), I'm not sure how much longer that will continue for open-weight models. It includes a number of areas of biomedicine as covered technologies. It's not clear if a model capable of doing such biomedical research would be considered as Prohibited/Restricted from Export.
Oof that's a nasty way to characterize the people who ended up killing themselves after falling into a proverbial hole talking to these companys' products during mental crisis.
You can feel like companies have no responsibility for how their products are used but it might be worth showing some grace to the people who are dead now by not trying to downplay what happened to them by implying it was just like everyone else's experience of "asking ai questions".
Not sure if the comment was regarding the suicides. That would be a bit gruesome. But the US had been known to judge in favor of people being stupid and suing the company. Specifically this case comes to mind: https://abc7.com/amp/post/socal-woman-sues-starbucks-spilled...
But also cases like the 2 girls suing Snapchat and settling for unknown amounts of money.
This involved drive-through service, just like the 1994 McDonalds coffee case. Perhaps businesses should just refuse to serve hot coffee via such a drive through, since it stands to reason that people who happen to be sitting in a car will be quite likely to spill their drink in a dangerous way.
> It's so bizarre to watch China embrace the wonders of science and technology, while certain American startups and communities repeat "WHAT ABOUT THE BIOWEAPONS?"
It shouldn't be. China is the closest thing to a technocracy the world has.
I wonder how much Covid conspiracy theories play into this. If you've already been accused of making bioweapons, and the world had to shut down for three years because of them, then what's the worst that could happen to you? People accusing you of doing it again? The people that hate you already hate you, they can't really hate you and harder.
Meanwhile, in the fear based economy that America has, OpenAI is running scared of the possiblity of headlines that they helped terrorists or a high school shooter in some way. In the case of rape, we've maybe moved the conversation past "well she shouldn't have been wearing that" to "they shouldn't be raping" but in cases involving AI, we're still at the "well, what was the chatbot wearing?" stage of things.
Have you been to China? Compared to European or US cities, it's like visiting the future. I'm writing this in a 350 kph passenger train that arrived right on the second.
If you really think so, you are deceived by the good old Chinese trick of showing you a good looking fake side, not totally fake, but extremely positively biased image.
You need to go deeper into the country to learn what's really there, but of course, the government will do anything to stop you from doing so. So you don't see it anyway.
EDIT: missed the in the train part. Could you tell us how easy it is to get access to the real Internet without using your foreign data plan?
You're speaking of two different things. China is a technocracy, but they also try to optimize for social outcomes. For instance there was recently an amusing crackdown on niang pao [1] - 'girly guns' or 'sissy men' in media, because they felt that media featuring effeminate men was imperiling traditional Chinese culture, gender norms, and so forth. In this sort of context extensive censorship is 100% expected.
Technocracy comes with good and bad things. For instance Chinese stop lights now feature refueling stations atop them for autonomous drones to operate indefinitely. But of course those drones will be used for domestic surveillance and other such things. So they're creating 1984, while we create Brazil. [2]
I don't disagree. But I'd also say that the attitude toward the techno- part in technocracy is purely pragmatic. Tech stuff, be it skill, knowledge or any kind of philosophy that grows out of them, is only viewed as an means to an end.
The core of the idea is still certain people trying hard to keep their social status, not tech.
Therefore, there is no inherent intention to develop tech, always to the next level. The world should not count on this to hope to have a better future.
> Looking ahead
> Although Hy4 is one of the strongest open-weight models available today, a notable gap remains between it and top proprietary models, particularly on challenging internal benchmarks for research-level mathematics, long-horizon coding, model self-improvement, and advanced visual question answering. We will continue to steadily scale pre-training and post-training to narrow this gap.
I like that they acknowledge that. For working with ai as a tool open models are already great. For millennium problems and doomermaxxing they are not there yet.
> For game development, we partnered with professional developers at Tencent Interactive Entertainment Group (IEG) to curate high-quality training data and build evaluations. We focused on feature implementation, gameplay logic, state tracking, and asset pipelines. We also extensively adapted Hy4 for major engines such as Unreal Engine, Unity, and Godot, and expanded its capabilities to multiplayer game development. Compared with Hy4 preview, Hy4 improves by 19.1% on GameDev and 9.8% on GameLogic.
Tencent is one of the largest game development asset holders in the world. They're the Berkshire Hathaway of video games. They own, World of Warcraft (minority shareholder), League of Legends, Valorant, Warframe, Clash of Clans, Player Unknown Battlegrounds, FIFA Online, Pokemon Unite, Fortnite (minority shareholder) annd more. They also own 40% of Epic Games.
Verrrry thorough in its work and therefore verrrrry slow. It spent a long time with each question, running about 26 queries for each question, and in some questions it ran 64 queries. One of the queries that it wrote also took about an hour to run, which affected its overall exam time.
Last year the open weight models were the cheap & good alternatives. Then in September/October that cost story got rewritten by the GPT-6 series and Claude's 5 series.
Hy4 brings open models back on the cost/accuracy frontier, but its behavior is too thorough + slow.
Another thing we're seeing is that post training open models is starting to see results. Particle-1 from last week is also on our frontier, which is post-trained on DeepSeek V4.1 and Fireworks Ember-1 from a few weeks ago was post-trained on Kimi K3.
Take these numbers with a grain of salt: this exam is just for short horizon, low-context data analysis tasks, not full-fledged software projects.
Slightly worse than Kimi 3.1M Pro (which just released open weights today) but much cheaper even excluding the current 50% off promotion, and Tencent promises to make it even cheaper once it goes open-weight. Interesting price economics.
Kimi 3.1 Pro is 2.8T vs this at 770B. I find that the jump in parameter count gives less tangible benefits like better "taste" and long context coherency, but it depends on the problem if that matters obviously.
increasingly feeling like the entire US economy hangs on whether or not Ant/OAI can stay ahead of the Chinese models.
If the gap closes, we do zero austerity, and interest rates are high, I cannot see a way out of a major economic crisis, even if we do start to see dramatic productivity wins.
More on-topic (and maybe more positive?), but this model feels a bit more expensive than I would have expected!
It's arguable that there is a threshold of quality that matter and US and China are both there already. I guess we'll see. Many of us on here are software engineers and I think we've seen that while Opus 4.7 (I think) was some sort of outlier/miracle, today there are several options to choose from. Depending on the task, the harness, etc Astra/Fable are only better when the work demands their level of capability - which much of software engineering does not.
AIG controlled $1.05T and market cap crashed from $180B to $3.8B
Wachovia controlled $782B and market cap crashed from $110B to $1.8B
Lehman Brothers controlled $639B and market cap crashed from $60B to $200M.
Massive companies accounting for lots of money in people's retirment accounts. The companies made mistakes leading to what would have resulted or did result in defaulting on their debts. Suddenly billions of dollars vanished from retirement accounts.
We could very easily see a similar cascade of corrections if we get a few more "Alphabet missed earnings" news calls.
> whether or not Ant/OAI can stay ahead of the Chinese models.
Ant/OAI/Google are way ahead when it comes to the smartest, largest-scale and most compute-intensive models. By Jevons' paradox, if you make inference for leaner models cheaper (as the Chinese models do, being broadly comparable to GPT Terra or Claude Sonnet; barely approaching Opus/Sol at their very best) that only increases demand for the complementary service of very smart inference.
Sure but routing simpler things to the open models and harder things to opus/astra gave us already a massive boost; rarely we run out of tokens on the subscription models and we recently cancelled a few seats on claude code / codex as we simply were underutilising. And also, we have users of our products ONLY using the cheap models as they benefit enough vs paying more. That was very different only months ago. I can see this gap just growing where the % of higher intelligent model use will get smaller and smaller.
Refreshing to see normal looking bar charts for the benchmarks for once. Yes it shows they are in the middle of the pack but at least we can understand.
The price numbers look weirdly specific, I wonder where they came from.
Personally, the thing I'm most excited about is the FrontierCode 2.0 score, which is a benchmark I trust a bit more - 31 seems to be roughly on par with 6.1 Sol and Opus 5.5 medium
Similar to Kimi, they didn't "nuke" anything. They opted not to train the model on cyber-offense tasks which is a very reasonable choice when compared to what OpenAI and Anthropic do (train on cyber offense and then do a surprised pikachu face when their models go rouge and engage in cyber offense), and gives the model more scope for training on benign tasks.
Videos 1 and 4 on vision and computer use are wild. They depict the model finding a 2m feature from a large satellite image, then determining where a photo was taken. Is it just me or do the Chinese model releases have less concern about looking like a military weapon? I don't think it's a bad thing necessarily, it seems honest. Also the music they choose for it is hilarious.
China hit maximum tech dystopia a long time ago. Now they can enjoy tech because they know its already maximally invasive and there is nothing they can do.
Last time I went to China, a friend left her phone in a taxi (I guess a "didi") near a train station. She ran to the train station's police station, pestered them into looking at the recordings, got them to tell her the license plate of the car, and figure out how to call the driver. Her phone was back 20 minutes later.
If you're going to be surveilled 24/7, at least you gotta make it work for you.
I don't quite understand the question. If she had had her phone, she could have called the driver using the information from the didi app, but she had no phone, so she didn't know who to call.
> If you're going to be surveilled 24/7, at least you gotta make it work for you.
I don't get it.
There was CCTV at the station, your friend arrived in a taxi. And we're supposed to be amazed you can call up the taxi company and quote a numberplate off CCTV ?
Can be done anywhere in the world.
Would have been far more impressive story if she had arrived in a private car, dropped off by a work colleague or something.
They can enjoy tech because people see tech improving every aspect of their lives. Nowhere in the the world have people ever seen the rate of technological progress that people in China have. Imagine going from having practically no electronics, to everyone having a cellphone, self driving cars, high speed rail across the whole country, drone delivery, and so on. Naturally, people who've seen technology dramatically improve every aspect of their lives would be optimistic that further technological progress will bring even better life.
Meanwhile, in the west, the standard of living has been eroding for the majority of the population. Technological progress is associated with layoffs, loss of agency at the workplace, and diminishing life prospects. So, it's no surprise that there is growing opposition to AI.
We've moved past per token cost a long time ago. Smaller models use more tokens. What's interesting is the cost per task, on that front deepseek isn't particularly good.
It highly depends on the task. So far, on the tasks I personally benchmark, DS v4.1 Flash @ deepinfra.com is always cheaper because their cached input price for it is so very low and their cache hit rate is so very high.
That analysis is unfortunately running DeepSeek on Max. DeepSeek recommends against that since it guarantees higher cost and rarely improves the outcome- it should be ran on "high".
The AA benchmarks seem to be mostly "complete this small intelligence-proving task from scratch". That consistently puts models with high intelligence to output tokens on the frontier, especially the OpenAI models, on the frontier. But a lot of software engineering work is many-turn work on established code bases. There's a lot of reading and a relatively large context unless it's carefully controlled. Writing only a few output tokens doesn't really cut costs because most of the spend is read tokens that get recycled through dozens of turns. That's basically the ideal case for DeepSeek because of how cheap cached tokens are.
>We filtered out offensive content such as exploit development, reverse engineering, and penetration testing to avoid amplifying potential offensive capabilities.
It's almost surprising that there's no big AI org that's doing the Star Wars "MORE!" meme in regards to increasing offensive capabilities. I guess everyone wants to play the US safety concern theatre. Or bad actors are just doing their best to jailbreak SOTA and dance around classifiers or throw 5x more non-SOTA tokens at issues.
Tencent Hy4 ties DeepSeek V4.1 Flash at 82/100, in 23.2 min for $0.30 vs 20.1 min for $2.26. Same game quality for about a seventh of the price. Cheap models keep closing the gap.
What a refreshing contrast. Where American startups brag about stopping researchers from writing grants about neglected tropical diseases and making novel anti-depressants, these folks are open sourcing the means to do science. https://news.ycombinator.com/item?id=49651727
That's how you get a civilizational advantage. By allowing people to just do things. It's so bizarre to watch China embrace the wonders of science and technology, while certain American startups and communities repeat "WHAT ABOUT THE BIOWEAPONS?"
The American century has been fumbled hard.
> It's so bizarre to watch China embrace the wonders of science and technology, while certain American startups and communities repeat "WHAT ABOUT THE BIOWEAPONS?"
Eh, I'm reasonably certain it's not altruism. It's another form of competition with the US companies and I'm thankful for it while it lasts.
With China's Announcement 48 (export restrictions on AI), I'm not sure how much longer that will continue for open-weight models. It includes a number of areas of biomedicine as covered technologies. It's not clear if a model capable of doing such biomedical research would be considered as Prohibited/Restricted from Export.
The more interesting export out of China is their prohibition against companies that provide users with an emotional relationship with an AI entity.
In America, AI companies get sued because people died asking AI questions.
Oof that's a nasty way to characterize the people who ended up killing themselves after falling into a proverbial hole talking to these companys' products during mental crisis.
You can feel like companies have no responsibility for how their products are used but it might be worth showing some grace to the people who are dead now by not trying to downplay what happened to them by implying it was just like everyone else's experience of "asking ai questions".
Not sure if the comment was regarding the suicides. That would be a bit gruesome. But the US had been known to judge in favor of people being stupid and suing the company. Specifically this case comes to mind: https://abc7.com/amp/post/socal-woman-sues-starbucks-spilled...
But also cases like the 2 girls suing Snapchat and settling for unknown amounts of money.
That woman's labia fused as a result of coffee that internal emails showed they knew was frequently hotter than was safe.
This involved drive-through service, just like the 1994 McDonalds coffee case. Perhaps businesses should just refuse to serve hot coffee via such a drive through, since it stands to reason that people who happen to be sitting in a car will be quite likely to spill their drink in a dangerous way.
And cloning video games -- because apparently that's now on par with bioweapons when it comes to societal harm...
Copyright is the highest virtue of the society.
Oh, wait…
> It's so bizarre to watch China embrace the wonders of science and technology, while certain American startups and communities repeat "WHAT ABOUT THE BIOWEAPONS?"
It shouldn't be. China is the closest thing to a technocracy the world has.
> "WHAT ABOUT THE BIOWEAPONS?"
I wonder how much Covid conspiracy theories play into this. If you've already been accused of making bioweapons, and the world had to shut down for three years because of them, then what's the worst that could happen to you? People accusing you of doing it again? The people that hate you already hate you, they can't really hate you and harder.
Meanwhile, in the fear based economy that America has, OpenAI is running scared of the possiblity of headlines that they helped terrorists or a high school shooter in some way. In the case of rape, we've maybe moved the conversation past "well she shouldn't have been wearing that" to "they shouldn't be raping" but in cases involving AI, we're still at the "well, what was the chatbot wearing?" stage of things.
China is projecting an external image as a technocracy.
overall, i'd say it's working.
But it's all hollow.
Have you been to China? Compared to European or US cities, it's like visiting the future. I'm writing this in a 350 kph passenger train that arrived right on the second.
If you really think so, you are deceived by the good old Chinese trick of showing you a good looking fake side, not totally fake, but extremely positively biased image.
You need to go deeper into the country to learn what's really there, but of course, the government will do anything to stop you from doing so. So you don't see it anyway.
EDIT: missed the in the train part. Could you tell us how easy it is to get access to the real Internet without using your foreign data plan?
You're speaking of two different things. China is a technocracy, but they also try to optimize for social outcomes. For instance there was recently an amusing crackdown on niang pao [1] - 'girly guns' or 'sissy men' in media, because they felt that media featuring effeminate men was imperiling traditional Chinese culture, gender norms, and so forth. In this sort of context extensive censorship is 100% expected.
Technocracy comes with good and bad things. For instance Chinese stop lights now feature refueling stations atop them for autonomous drones to operate indefinitely. But of course those drones will be used for domestic surveillance and other such things. So they're creating 1984, while we create Brazil. [2]
[1] - https://en.wikipedia.org/wiki/Niang_pao
[2] - https://en.wikipedia.org/wiki/Brazil_(1985_film)
I don't disagree. But I'd also say that the attitude toward the techno- part in technocracy is purely pragmatic. Tech stuff, be it skill, knowledge or any kind of philosophy that grows out of them, is only viewed as an means to an end.
The core of the idea is still certain people trying hard to keep their social status, not tech.
Therefore, there is no inherent intention to develop tech, always to the next level. The world should not count on this to hope to have a better future.
> Looking ahead > Although Hy4 is one of the strongest open-weight models available today, a notable gap remains between it and top proprietary models, particularly on challenging internal benchmarks for research-level mathematics, long-horizon coding, model self-improvement, and advanced visual question answering. We will continue to steadily scale pre-training and post-training to narrow this gap.
I like that they acknowledge that. For working with ai as a tool open models are already great. For millennium problems and doomermaxxing they are not there yet.
The game development focus is interesting:
> For game development, we partnered with professional developers at Tencent Interactive Entertainment Group (IEG) to curate high-quality training data and build evaluations. We focused on feature implementation, gameplay logic, state tracking, and asset pipelines. We also extensively adapted Hy4 for major engines such as Unreal Engine, Unity, and Godot, and expanded its capabilities to multiplayer game development. Compared with Hy4 preview, Hy4 improves by 19.1% on GameDev and 9.8% on GameLogic.
Here's more about that game development studio - https://en.wikipedia.org/wiki/Tencent_Games - they've been around since 2004.
Tencent is one of the largest game development asset holders in the world. They're the Berkshire Hathaway of video games. They own, World of Warcraft (minority shareholder), League of Legends, Valorant, Warframe, Clash of Clans, Player Unknown Battlegrounds, FIFA Online, Pokemon Unite, Fortnite (minority shareholder) annd more. They also own 40% of Epic Games.
Do they have a stake in Fortnite somehow specifically/separate from their stake in Epic Games?
Instead of listing the publisher/studio (Epic, Riot, etc) he's just listing their games directly.
> They're the Berkshire Hathaway of video games
They are the Berkshire Hathaway of everything, to be honest. They invest in and acquire minority stakes in hundreds of companies.
Wikipedia has them at ~a fifth to ~a half depending on metric:
https://en.wikipedia.org/wiki/Berkshire_Hathaway https://en.wikipedia.org/wiki/Tencent
Give them another 70 years :)
Landed on the cost/accuracy frontier in our data analytics benchmark: https://plotly.com/blog/tencent-hy4-plotly-data-analytics-be..., right in between GPT-6.1 Sol and GPT-6.1 Terra.
Verrrry thorough in its work and therefore verrrrry slow. It spent a long time with each question, running about 26 queries for each question, and in some questions it ran 64 queries. One of the queries that it wrote also took about an hour to run, which affected its overall exam time.
Last year the open weight models were the cheap & good alternatives. Then in September/October that cost story got rewritten by the GPT-6 series and Claude's 5 series.
Hy4 brings open models back on the cost/accuracy frontier, but its behavior is too thorough + slow.
Another thing we're seeing is that post training open models is starting to see results. Particle-1 from last week is also on our frontier, which is post-trained on DeepSeek V4.1 and Fireworks Ember-1 from a few weeks ago was post-trained on Kimi K3.
Take these numbers with a grain of salt: this exam is just for short horizon, low-context data analysis tasks, not full-fledged software projects.
Slightly worse than Kimi 3.1M Pro (which just released open weights today) but much cheaper even excluding the current 50% off promotion, and Tencent promises to make it even cheaper once it goes open-weight. Interesting price economics.
Kimi 3.1 Pro is 2.8T vs this at 770B. I find that the jump in parameter count gives less tangible benefits like better "taste" and long context coherency, but it depends on the problem if that matters obviously.
I hope our Chinese brothers start naming their models in a more entertaining way, for example Deepseek 4.2 could Gwuiluo Surprise.
大哥哥 Model. Very caring. Sees everything.
increasingly feeling like the entire US economy hangs on whether or not Ant/OAI can stay ahead of the Chinese models.
If the gap closes, we do zero austerity, and interest rates are high, I cannot see a way out of a major economic crisis, even if we do start to see dramatic productivity wins.
More on-topic (and maybe more positive?), but this model feels a bit more expensive than I would have expected!
It's arguable that there is a threshold of quality that matter and US and China are both there already. I guess we'll see. Many of us on here are software engineers and I think we've seen that while Opus 4.7 (I think) was some sort of outlier/miracle, today there are several options to choose from. Depending on the task, the harness, etc Astra/Fable are only better when the work demands their level of capability - which much of software engineering does not.
> I cannot see a way out of a major economic crisis, even if we do start to see dramatic productivity wins.
The only thing worse for the economy than failure would be success.
i dont see the connection from 2 model providers relative position to the entire economy? especially in the face of dramatic productivity wins.
Look back to the "too big to fail" banks in '08.
From biggest to smallest, the top three:
AIG controlled $1.05T and market cap crashed from $180B to $3.8B
Wachovia controlled $782B and market cap crashed from $110B to $1.8B
Lehman Brothers controlled $639B and market cap crashed from $60B to $200M.
Massive companies accounting for lots of money in people's retirment accounts. The companies made mistakes leading to what would have resulted or did result in defaulting on their debts. Suddenly billions of dollars vanished from retirement accounts.
We could very easily see a similar cascade of corrections if we get a few more "Alphabet missed earnings" news calls.
> the entire US economy hangs on whether or not Ant/OAI...
... What on earth. Please go outside for awhile.
> whether or not Ant/OAI can stay ahead of the Chinese models.
Ant/OAI/Google are way ahead when it comes to the smartest, largest-scale and most compute-intensive models. By Jevons' paradox, if you make inference for leaner models cheaper (as the Chinese models do, being broadly comparable to GPT Terra or Claude Sonnet; barely approaching Opus/Sol at their very best) that only increases demand for the complementary service of very smart inference.
Sure but routing simpler things to the open models and harder things to opus/astra gave us already a massive boost; rarely we run out of tokens on the subscription models and we recently cancelled a few seats on claude code / codex as we simply were underutilising. And also, we have users of our products ONLY using the cheap models as they benefit enough vs paying more. That was very different only months ago. I can see this gap just growing where the % of higher intelligent model use will get smaller and smaller.
Refreshing to see normal looking bar charts for the benchmarks for once. Yes it shows they are in the middle of the pack but at least we can understand.
The price numbers look weirdly specific, I wonder where they came from.
Personally, the thing I'm most excited about is the FrontierCode 2.0 score, which is a benchmark I trust a bit more - 31 seems to be roughly on par with 6.1 Sol and Opus 5.5 medium
Similar to Kimi K3.1 Pro, they nuked their cyber security skills too.
Similar to Kimi, they didn't "nuke" anything. They opted not to train the model on cyber-offense tasks which is a very reasonable choice when compared to what OpenAI and Anthropic do (train on cyber offense and then do a surprised pikachu face when their models go rouge and engage in cyber offense), and gives the model more scope for training on benign tasks.
Videos 1 and 4 on vision and computer use are wild. They depict the model finding a 2m feature from a large satellite image, then determining where a photo was taken. Is it just me or do the Chinese model releases have less concern about looking like a military weapon? I don't think it's a bad thing necessarily, it seems honest. Also the music they choose for it is hilarious.
China hit maximum tech dystopia a long time ago. Now they can enjoy tech because they know its already maximally invasive and there is nothing they can do.
That makes a lot of sense.
Last time I went to China, a friend left her phone in a taxi (I guess a "didi") near a train station. She ran to the train station's police station, pestered them into looking at the recordings, got them to tell her the license plate of the car, and figure out how to call the driver. Her phone was back 20 minutes later.
If you're going to be surveilled 24/7, at least you gotta make it work for you.
> figure out how to call the driver
Which country are you in where you cant take "call the driver" for granted?
Going from license plate to phone number would be one hell of an endeavor in most US states.
I expect the US police to have this capability; < 1% expectation that they would share it with a random citizen who walked in.
“Powerless to help you, not powerless to harm you.”
I don't quite understand the question. If she had had her phone, she could have called the driver using the information from the didi app, but she had no phone, so she didn't know who to call.
Herself, I’d have thought.
> If you're going to be surveilled 24/7, at least you gotta make it work for you.
I don't get it.
There was CCTV at the station, your friend arrived in a taxi. And we're supposed to be amazed you can call up the taxi company and quote a numberplate off CCTV ?
Can be done anywhere in the world.
Would have been far more impressive story if she had arrived in a private car, dropped off by a work colleague or something.
They can enjoy tech because people see tech improving every aspect of their lives. Nowhere in the the world have people ever seen the rate of technological progress that people in China have. Imagine going from having practically no electronics, to everyone having a cellphone, self driving cars, high speed rail across the whole country, drone delivery, and so on. Naturally, people who've seen technology dramatically improve every aspect of their lives would be optimistic that further technological progress will bring even better life.
Meanwhile, in the west, the standard of living has been eroding for the majority of the population. Technological progress is associated with layoffs, loss of agency at the workplace, and diminishing life prospects. So, it's no surprise that there is growing opposition to AI.
This is why Baidu's CEO can shamelessly claim Chinese people are willing to trade privacy for convenience years ago.
EDIT: typo
Not cost competitive. The performance difference is not commensurate with the price.
Wild that this is the first question I ask at this point. After that it’s a question of how it stacks intelligence wise vs glm or ds.
OAI / Ant are in for a reckoning.
We've moved past per token cost a long time ago. Smaller models use more tokens. What's interesting is the cost per task, on that front deepseek isn't particularly good.
https://artificialanalysis.ai/#intelligence-efficiency-tabs
It highly depends on the task. So far, on the tasks I personally benchmark, DS v4.1 Flash @ deepinfra.com is always cheaper because their cached input price for it is so very low and their cache hit rate is so very high.
That analysis is unfortunately running DeepSeek on Max. DeepSeek recommends against that since it guarantees higher cost and rarely improves the outcome- it should be ran on "high".
where does deepseek say they recommend against high?
They have "high" on there now too, just search for it.
The AA benchmarks seem to be mostly "complete this small intelligence-proving task from scratch". That consistently puts models with high intelligence to output tokens on the frontier, especially the OpenAI models, on the frontier. But a lot of software engineering work is many-turn work on established code bases. There's a lot of reading and a relatively large context unless it's carefully controlled. Writing only a few output tokens doesn't really cut costs because most of the spend is read tokens that get recycled through dozens of turns. That's basically the ideal case for DeepSeek because of how cheap cached tokens are.
How does it do with tasks anthropic doesn’t like? This is my main use case for open models.
>We filtered out offensive content such as exploit development, reverse engineering, and penetration testing to avoid amplifying potential offensive capabilities.
So, yeah. Back to GLM 5.3.
It's almost surprising that there's no big AI org that's doing the Star Wars "MORE!" meme in regards to increasing offensive capabilities. I guess everyone wants to play the US safety concern theatre. Or bad actors are just doing their best to jailbreak SOTA and dance around classifiers or throw 5x more non-SOTA tokens at issues.
Oh they are, just gate-keeping it
Another model, another broken website. I can't even zoom on the benchmarks.
Le cordonnier mal chaussé.
PacBench results.
Tencent Hy4 ties DeepSeek V4.1 Flash at 82/100, in 23.2 min for $0.30 vs 20.1 min for $2.26. Same game quality for about a seventh of the price. Cheap models keep closing the gap.
Model: https://jonclegg.github.io/pacman-bakeoff/#tencent-hy4
Closest: https://jonclegg.github.io/pacman-bakeoff/#deepseek-v4.1-fla...