>The same dynamic runs through the open evals, smaller models that feel dumber in practice still outscore better ones on sites like Artificial Analysis. This isn’t hypothetical, Thinking Machines’ Inkling Small scored within a point of its flagship sibling on the Artificial Analysis Intelligence Index with less than a third of the parameters and beat it on Humanity’s Last Exam, GPQA Diamond and SciCode. Public, static, famous test sets leak into training data and fine-tuning choices.
This sounds unconvincing, because a) pelican test is subjective, there's simply nothing to leak as it has no available direct answers and maybe an extremely faint preference signal, and b) the same small models actually do perform well when you change the subject. Some models are genuinely trained to be better at some domain, such as 2D layouts or vector graphics in this case. It all depends on particular recipes and datasets. This is the actual reason these tests are poor as vibe checks: they don't do anything to disentangle generalization, memorization, and specialized training.
Build a flowsheet based steady state chemical process simulator, then use it to simulate and optimize a full scale oil refinery.
1) Building a solver engine that works at this scale is not a trivial problem, and the successful ones rely more on heuristics than some categorically different solution approach.
2) Defining the engineering equations relevant to this task is relies on understanding what level of fidelity is required to answer the questions people ask of steady state process models.
3) Knowing the thermophysical properties of chemicals and crude oils is possible from the open literature, but the information is diffuse and different correlations are applicable in different situations.
4) Creating a GUI which converts a flowsheet into matrix math is non-trivial, although a sequential modular approach is a bit easier.
5) Defining large scale models in such a way that they solve robustly is as much art as science. For example, completely closed recycle loops like refrigeration systems are a nightmare for solvers, so it is often better to define them in an open-loop way.
6) Optimization involves knowing the relevant commodity prices, but more importantly how to define the constraints on the model so it doesn't just say to produce infinite gasoline.
7) Troubleshooting the inevitable convergence failures is also as much art as science. There are a large number of diagnostic techniques, but fundamentally you need to be able to relate what is happening during the solver iterations with the intent of your model because more often than not the problem is that you've asserted something impossible, redundant, or irrelevant.
I keep seeing Astra make beautiful 3d stuff online, yet when I feed it some old school RuneScape assets (even tried with some very detailed guidelines) and asked it to generate some new plausible assets it failed horribly.
I think there’s still something really off with current (frontier) models when it comes to creating “novel” stuff? Even 2004 style graphics..
Though I somewhat think some benchmarks are silly like the article says. I saw someone on YouTube recently take a picture of a building across the street from them (seemed like it was in NYC), and asked GPT 6 Astra to make it in blender. It did a surprisingly good job in 30 minutes. So though these benchmarks don't seem to mean much you could always add a touch of randomness to them like the person in the YouTube video did, but the problem with that is how would you compare the benchmarks in any clear way if they aren't even consistent? Regardless it appears LLMs are getting this good at the general task and not just at the particular instances of said task.
> That’s the problem, these tests can’t tell you how good a model is anymore because it’s trivial for labs to optimise for exactly these tests by the next release.
But they don't prove the claim. Are the models amazing at recreating Minecraft, but the second you swap the word Minecraft out with another game or a custom game, it shits the bed?
That's not what I see. My feed is full of people using Astra to recreate all sorts of games from Diablo to some random idea they came up with, in ridiculously polished detail like animations that would have taken me weeks of iteration in gpt-5.6-sol but it was a single shot by Astra.
Can you point me to one of these? The only Astra game I tried was the incredibly janky Mario Kart clone from OpenAI where you could fly by spamming jump
yeah, that's a good point. In a followup Anshu is saying he ran Astra for 8 hours to overhaul that game.. isn't that going to be ungodly expensive just for a test of the model?
For every person using Astra to create a game from some random idea and sharing the impressive result, there are an unknown number who have tried the same thing and gave up in frustration.
@notch Hey Markus, can I get a refund on my alpha distro of minecraft? I think the currency is worth more than it used to be considering how many versions there are now :)
Running Firefox here, no problems whatsoever even with UBO and Privacy Badger disabled (the twitter embeds get blocked at the DNS level, but I doubt those are the problem).
Mac here. There's nothing all that interesting going on with the JS on that page, so I would suspect a bug with Firefox and/or your desktop environment.
I’ve used Astra for the past day and a half. My layperson’s review is that it is impressive at computer use and 3D reasoning, and fails in similar ways to 5.6 Sol at similar rates when it comes to coding. I have no idea how it scored so high on SWE benchmarks because so far it has been very “mid” as the kids say.
Most of my and my peers PortCos run their own eval and benchmark sets, simply because they know what they need best.
The reality is, capabilities have largely converged across foundation models over the last 18 months, and much of the value add is coming from the harness layer itself now.
This has been the operating assumption for me and my peers, and has largely played out that way.
That said, this has always been an issue with benchmarking since the very beginning. DB Benchmarks, compute benchmarks, and others that were external facing were always inherently a content and product marketing tool. The actual internal benchmarking used to model, understand, and enhance your product was always a closely held secret.
Most of these conversations are happening, but largely in person and not on HN.
>The same dynamic runs through the open evals, smaller models that feel dumber in practice still outscore better ones on sites like Artificial Analysis. This isn’t hypothetical, Thinking Machines’ Inkling Small scored within a point of its flagship sibling on the Artificial Analysis Intelligence Index with less than a third of the parameters and beat it on Humanity’s Last Exam, GPQA Diamond and SciCode. Public, static, famous test sets leak into training data and fine-tuning choices.
This sounds unconvincing, because a) pelican test is subjective, there's simply nothing to leak as it has no available direct answers and maybe an extremely faint preference signal, and b) the same small models actually do perform well when you change the subject. Some models are genuinely trained to be better at some domain, such as 2D layouts or vector graphics in this case. It all depends on particular recipes and datasets. This is the actual reason these tests are poor as vibe checks: they don't do anything to disentangle generalization, memorization, and specialized training.
The benchmark I want to see people adopt is:
Build a flowsheet based steady state chemical process simulator, then use it to simulate and optimize a full scale oil refinery.
1) Building a solver engine that works at this scale is not a trivial problem, and the successful ones rely more on heuristics than some categorically different solution approach.
2) Defining the engineering equations relevant to this task is relies on understanding what level of fidelity is required to answer the questions people ask of steady state process models.
3) Knowing the thermophysical properties of chemicals and crude oils is possible from the open literature, but the information is diffuse and different correlations are applicable in different situations.
4) Creating a GUI which converts a flowsheet into matrix math is non-trivial, although a sequential modular approach is a bit easier.
5) Defining large scale models in such a way that they solve robustly is as much art as science. For example, completely closed recycle loops like refrigeration systems are a nightmare for solvers, so it is often better to define them in an open-loop way.
6) Optimization involves knowing the relevant commodity prices, but more importantly how to define the constraints on the model so it doesn't just say to produce infinite gasoline.
7) Troubleshooting the inevitable convergence failures is also as much art as science. There are a large number of diagnostic techniques, but fundamentally you need to be able to relate what is happening during the solver iterations with the intent of your model because more often than not the problem is that you've asserted something impossible, redundant, or irrelevant.
I keep seeing Astra make beautiful 3d stuff online, yet when I feed it some old school RuneScape assets (even tried with some very detailed guidelines) and asked it to generate some new plausible assets it failed horribly.
I think there’s still something really off with current (frontier) models when it comes to creating “novel” stuff? Even 2004 style graphics..
Or am promoting it wrong?
Did you try blender MCP and prompt for low-poly vertex colored assets?
Though I somewhat think some benchmarks are silly like the article says. I saw someone on YouTube recently take a picture of a building across the street from them (seemed like it was in NYC), and asked GPT 6 Astra to make it in blender. It did a surprisingly good job in 30 minutes. So though these benchmarks don't seem to mean much you could always add a touch of randomness to them like the person in the YouTube video did, but the problem with that is how would you compare the benchmarks in any clear way if they aren't even consistent? Regardless it appears LLMs are getting this good at the general task and not just at the particular instances of said task.
> That’s the problem, these tests can’t tell you how good a model is anymore because it’s trivial for labs to optimise for exactly these tests by the next release.
But they don't prove the claim. Are the models amazing at recreating Minecraft, but the second you swap the word Minecraft out with another game or a custom game, it shits the bed?
That's not what I see. My feed is full of people using Astra to recreate all sorts of games from Diablo to some random idea they came up with, in ridiculously polished detail like animations that would have taken me weeks of iteration in gpt-5.6-sol but it was a single shot by Astra.
> ridiculously polished detail
Can you point me to one of these? The only Astra game I tried was the incredibly janky Mario Kart clone from OpenAI where you could fly by spamming jump
X.com is full of unverified posts, but here's a few:
- https://x.com/MengTo/status/2096213835460084184?s=20
- https://x.com/birdabo/status/2096101381925114327?s=20
Seems really impressive at first glance, but there's no guarantee that they are real
> no guarantee that they are real
yeah, that's a good point. In a followup Anshu is saying he ran Astra for 8 hours to overhaul that game.. isn't that going to be ungodly expensive just for a test of the model?
You are falling for selection bias.
For every person using Astra to create a game from some random idea and sharing the impressive result, there are an unknown number who have tried the same thing and gave up in frustration.
I call bullshit
@notch Hey Markus, can I get a refund on my alpha distro of minecraft? I think the currency is worth more than it used to be considering how many versions there are now :)
I cant select text nor click links on this page with firefox.
Running Firefox here, no problems whatsoever even with UBO and Privacy Badger disabled (the twitter embeds get blocked at the DNS level, but I doubt those are the problem).
Works on phone but not on my linux desktop.
Mac here. There's nothing all that interesting going on with the JS on that page, so I would suspect a bug with Firefox and/or your desktop environment.
Who cares? Restart your browser maybe
I’ve used Astra for the past day and a half. My layperson’s review is that it is impressive at computer use and 3D reasoning, and fails in similar ways to 5.6 Sol at similar rates when it comes to coding. I have no idea how it scored so high on SWE benchmarks because so far it has been very “mid” as the kids say.
Most of my and my peers PortCos run their own eval and benchmark sets, simply because they know what they need best.
The reality is, capabilities have largely converged across foundation models over the last 18 months, and much of the value add is coming from the harness layer itself now.
This has been the operating assumption for me and my peers, and has largely played out that way.
That said, this has always been an issue with benchmarking since the very beginning. DB Benchmarks, compute benchmarks, and others that were external facing were always inherently a content and product marketing tool. The actual internal benchmarking used to model, understand, and enhance your product was always a closely held secret.
Most of these conversations are happening, but largely in person and not on HN.