Wow this is interesting to me from another angle. A friend of mine created a benchmark to evaluate LLMs on 9x9 Go (using KataGo opponents as Elo anchors) [1]. The relative rankings of this (StarSkirmish Bench v0.2) benchmark (GPT-6 Astra > GPT-6.1 Sol > Gemini 4 Argon > Claude Opus 5.5 > Claude Fable 5.1 > Claude Sonnet 5.5) matches the order of his GoBench rankings exactly, even though they are different problems at surface level.
Is there anything interesting to learn from the agent written bots? Or are they just inferior versions of the best bots because the time limit is so short? Is there any indication that they’re trained on the code from the human written bots (like, do they magically “guess” creating functionality that requires experimental feedback or domain experience?) How much are the bots on rails applying the same techniques every time?
I think those kinds of questions would be more interesting than the raw win/loss ratios.
none of these games really require a lot of experimental feedback, the goals are well stated and known; experiments may drive optimization if thats the question.
as far as the on rails comment : there's no good way to know the provenance of an LLM output, but the LLM made bots generally look like very small, modular, sometimes functionalized versions of the human equivalents; the difference being that the LLM is calling the funtions often rather than a conditional loop in a main().
a factorio bot I worked on when Astra was first getting previewed was essentially a directory of very small argument-receiving functional and self-supporting programs that had very limited scope, and an agents.md file that directed the LLM to where such capabilities existed to preempt a directory search every compaction/context-swap.
so you'd have things like "python walk.py N 40" for 40 units of northern travel. a watchdog that waited for HP to drop or movement to become inhibited, etc.
My very first experience with LLM driven bot development in the GPT3 days created huge unmaintainmable monoliths that kind of worked due to TDD but were essentially spaghetti. LLMs have been trending away from those paradigms into small self-encapsulated programs. probably because those are less abusive to ctx count and produce better results since they lose/compact less.
This kind of makes me think about making better bots for Counter Strike based on human top level players, which i think its a very undervalued category of software. You can even forget about about CS and name any FPS and you'll be hard pressed to find a game with seriously good bots.
I am not sure I like this benchmark, but mostly, because I am not sure what it is really evaluating. It is not an idle dismissal.. multiple factors seem to complicate what it is supposed to look at ( just top of my head ):
- code generation speed
- logic analysis
- understanding of the game platform and how use of C may influence it
- considered approaches, tactics and strategies
Don't get me wrong. It is is neat idea. I am just not sure what it measures.
Yeah giving a 1 hour limit for code generation makes the whole thing a bit incoherent. Plus, performance is just going to depend on how closely the models have memorized the best open-source Starcraft bots and can reproduce them.
All of the above, to the extent that they can be measured from a simple to verify end condition (stats, win rates, etc). Adaptability, ability to understand and progress on a task, and so on.
I noticed Pluto is running on a simple CPU? If so, that's very nice for future RTS games. You just ship (or use a server-client arch) a bunch of models of various difficulty levels to let players practice. Could we even get models tuned to specific campaign levels? I wonder how costly the creation of Pluto was.
I assume lots of bot code is part of the training data? So it's basically a version of how will they can improve upon those solutions public out there?
Please don't comment on whether someone read an article. "Did you even read the article? It mentions that" can be shortened to "The article mentions that".
How do they avoid the LLMs simply working with pre-training data? Which apparently they do, since the human written bots win. But it's a mystery, is what it is.
It's beyond insane that Astra achieves above 50%. Are models allowed to "play against" the specific bot they're up against? What explains the fact that it drops to single-digits against strong bots?
Yes during the 1 hour window they are allowed to play practice matches against opponents they are graded against - though they do not have access to their opponents source code.
They drop to single digits against the A tier and S tier bots because those bots are just generally harder opponents to beat.
If you are talking about take a screen shot, analyze move the window a bit, take another screenshot, move the window, repeat until they have seen enough of the map to make a decision, then decide what to click on, issue order.... this could absolutely be done. It's very similar to how LLM's have driven a car on a safe track.
But right now the response time would be so slow that the bot would not be close to competitive except with the most low level players. Especially because its hard to click on things that are moving at the same time as well.
Sometime soon, I'd like to see an LLM be competitive using only computer vision, mouse movement, and locked into a human possible number of actions per minute.
Even the most powerful frontier models on SOTA hardware are not fast enough for starcraft, which generally involves multiple informed decisions per second at peak moments.
You can train a specialized model for this, but a frontier model is ill-fitting.
There is currently an AI bot playing on the ladder in Korea and essentially it will nearly always win because it can micro every unit individually across the entire map. If a normal human can max out at hundreds of actions per minute, the bot can do thousands. Broodwar is so insanely difficult mechanically that there is no way for humans to compete. I imagine in a few months, maybe a year, it will be completely unbeatable.
Maybe because Protosses have been doing bad lately and need some help improving :) (won only 3 out of 22 ASL championships in the past 10 years)
On a more serious note: all three races play quite differently so it makes sense to focus on just one. And protoss is mechanically the easiest and most straightforward race: fewer but stronger units, a single worker can be used to build multiple buildings at once, versatile basic static defence
[1] https://rolandgao.com/blog/gobench/
I think those kinds of questions would be more interesting than the raw win/loss ratios.
as far as the on rails comment : there's no good way to know the provenance of an LLM output, but the LLM made bots generally look like very small, modular, sometimes functionalized versions of the human equivalents; the difference being that the LLM is calling the funtions often rather than a conditional loop in a main().
a factorio bot I worked on when Astra was first getting previewed was essentially a directory of very small argument-receiving functional and self-supporting programs that had very limited scope, and an agents.md file that directed the LLM to where such capabilities existed to preempt a directory search every compaction/context-swap.
so you'd have things like "python walk.py N 40" for 40 units of northern travel. a watchdog that waited for HP to drop or movement to become inhibited, etc.
My very first experience with LLM driven bot development in the GPT3 days created huge unmaintainmable monoliths that kind of worked due to TDD but were essentially spaghetti. LLMs have been trending away from those paradigms into small self-encapsulated programs. probably because those are less abusive to ctx count and produce better results since they lose/compact less.
- code generation speed - logic analysis - understanding of the game platform and how use of C may influence it - considered approaches, tactics and strategies
Don't get me wrong. It is is neat idea. I am just not sure what it measures.
All of the above, to the extent that they can be measured from a simple to verify end condition (stats, win rates, etc). Adaptability, ability to understand and progress on a task, and so on.
The customer may not care what the internal contributions are to such a measure. Instead it represents the customers’ overall values in some way.
I think with the abilities of AI improving, playing games would be a good benchmark as long as they're not trained on them.
It would be interesting to see the reverse, how long does each model take to get >50% WR against bots of various tiers.
Although I should caveat the format advantaged the LLM bots compared to traditional tournament rules.
They drop to single digits against the A tier and S tier bots because those bots are just generally harder opponents to beat.
TLDR they suck and many default to doing probe rushes.
But right now the response time would be so slow that the bot would not be close to competitive except with the most low level players. Especially because its hard to click on things that are moving at the same time as well.
Sometime soon, I'd like to see an LLM be competitive using only computer vision, mouse movement, and locked into a human possible number of actions per minute.
You can train a specialized model for this, but a frontier model is ill-fitting.
On a more serious note: all three races play quite differently so it makes sense to focus on just one. And protoss is mechanically the easiest and most straightforward race: fewer but stronger units, a single worker can be used to build multiple buildings at once, versatile basic static defence