On September 19 2026 16:12 fLyiNgDroNe wrote: anyone has a link to a vod from larva's stream when he played pluto? I can only imagine his reaction when he is getting trapped in terrans base by pure marines and cannot scale. This just never happens.
thanks mate. Larva's reaction when flying into the nat of terran "huh? no turrets?" lmao Terrans nowadays have to build at least 8 turrets to stay alive, thats 600 minerals in turret cost + lost mining time for 8 workers. Thats a lot of early economy. Pumping out marines non-stop from 4-5 barracks give you so much push power, that zerg needs to invest in the defense themselves. Its a completely different dynamic of the early game where instead of turtling terran is pushing. I don;t think this is humanly possible though. Terran needs to route marines in base perfectly, constantly re-target, stim pack each marine individually at exactly the right moment and anticipate movement.
if flash didn't have hand issues and had soma's ability to practice 50 games in a row in a single sitting, i can almost guarantee he would make turretless MM TvZ into a legit build
On September 19 2026 09:21 Crimson)S(hadow wrote: i don't see the bot anymore, did blizzard do something or is it on a new smurf?
^333^ that account seems removed... it would matter if ladder was working lel
HOLYBATS Profile PM Joined August 2021 Turkey841 Posts 4 hours ago #55 Maphack with 6k apm lol.
pretty much..
There is insane potential for improvement tho... strategy, building placement and even macro [idle barracks and factories with money to spare] and not using lockdown on self
HOLYBATS Profile PM Joined August 2021 Turkey841 Posts 4 hours ago #55 Maphack with 6k apm lol.
pretty much..
is ladder still down? soma and the proleague gang were having server issues last night, but it seemed to be fine once they launched sc through 'americas'
On September 19 2026 15:04 badbeatpete wrote: Wait so all this time the answer to 2hatch muta was to skip turrets completely and just get stim+range asap? Is this going to cause an immediate meta shift in TvZ?
Don't agree with that based on those larva games v bot being on TR8, need to see on regular TR24 to make judgement calls.TR8 is a hindrance to the human players here but no difference to the bot.
On September 19 2026 15:04 badbeatpete wrote: Wait so all this time the answer to 2hatch muta was to skip turrets completely and just get stim+range asap? Is this going to cause an immediate meta shift in TvZ?
Don't agree with that based on those larva games v bot being on TR8, need to see on regular TR24 to make judgement calls.TR8 is a hindrance to the human players here but no difference to the bot.
True, I did notice a lot of missed mutalisk shots from the zergs in a number of replays. Did not know about the turnrate issue. Knowing the AI is trained against itself, it probably wouldn't have learned stacked muta micro and thus developed the turretless style that is only viable against unclumped mutas. If turretless is actually viable, you'd think human players would have figured it out a long time ago.
I guess we do need to see a top zerg pick it apart on turnrate 24 to have confirmation though.
Yeah the bot is trained for bot vs. bot tournaments, which run on very low latency (corresponding to 1.16.1 LAN on Normal speed). So it is likely that whoever put the bot on the ladder wanted to force it to run at as low latency settings as possible so that it would perform better, which of course is an extra burden on the humans playing it. In the showmatches that NiteKat hosted, it was run at human-friendly turn rates and still played fine though.
This is another reason why we would prefer the bot to be made available for local offline play, where latency wouldn't be such an issue.
I'll also just throw in some more context for the APM discussion.
For me the really impressive thing with Pluto is how it compares to the other BW bots. Every other bot is multi-agent, meaning we independently perform actions on whatever units we want on every frame. So on the same frame a bot might micro a bunch of army units in a fight, issue some orders to workers to mineral boost, and command some production buildings to queue more units. Because of how its model works, Pluto is single-agent and only runs every 6 frames, so it can only decide to do one "thing" (virtually "selecting" or commanding what is currently selected) at a much slower rate compared to other bots.
So when you see Pluto playing against my bot, my bot will be maintaining something like 50k+ APM late game, and Pluto will be puttering along at 1000 while completely dominating. In that comparison, Pluto is much more "humanlike" since it is winning by making smarter decisions rather than by brute force.
Now going one step further and comparing it to actual humans gets into really muddy waters. Being single-agent, Pluto is still limited by stuff like needing to select buildings and queue them for macroing, for example (though I'm not sure if the model is allowed to select multiple buildings at the same time). But it doesn't need to spend any APM getting information, like moving the screen, clicking units to see how much health they have, etc. And it's also not limited by screen size when microing an army, which is something Artosis pointed out in the video. So it's very difficult for me to see what it is doing and conclude anything about whether a human could viably do it.
Ok saw Artosis's video and the Larva game, quite crazy even though it has maphack. It's breaking decade long traditions by sheer APM, micro and macro. No problem going mnm vs Protoss, it can target down your HTs and micro from 2 big screens before you get a big number of storms off. No turrets vs Zerg, no problem as well, it can pump micro marines in a circle perfectly.
On September 19 2026 19:14 Crimson)S(hadow wrote: bots don't need to stack mutas, in fact is is optimal to not stack them because it avoids splash https://www.youtube.com/watch?v=nNpy8Wnvf7g
Nope. What is optimal is that muta don't get hit at all.
On September 19 2026 16:37 Crimson)S(hadow wrote: i don't think its impossible for humans to execute something close, just give flash a new pair of hands
I don’t think so.
For it to work you need to essentially micro marines one by one so that they stay spread out enough to avoid getting massacred by storms and reavers, and to micro tanks so that they take out templars and reavers in priority.
While, obviously, macroing or of a trillion barracks perfectly.
Mm absolutely melt zealots and goons, so there is very little protoss can actually do. What is so striking in that game v err rsus Paralyze is that pluto won off two bases, because this army comp with that level of micro doesn’t even need a strong economic engine.
why does it matter if it uses map hack or not? i mean even if it does - still impressive, i remember checking those bots like 1-2 years ago and they were mediocre, this thing is on another level?
On September 19 2026 20:18 Zografa wrote: why does it matter if it uses map hack or not? i mean even if it does - still impressive, i remember checking those bots like 1-2 years ago and they were mediocre, this thing is on another level?
Yeah, the sad part for me is that I expect a lot of people discount it because of the maphack, while that actually isn't playing any role in how good it is. So it kind of spoiled the possibility to give it a proper reveal.
On September 19 2026 18:44 Vasoline73 wrote: Some of the Paralyze games were insane lmao. Opening Barracks, gas, acad, into expansion... fucking wild...
Maybe barracks, acadamy and ghost on lock down is the new meta.
On September 19 2026 16:37 Crimson)S(hadow wrote: i don't think its impossible for humans to execute something close, just give flash a new pair of hands
I don’t think so.
For it to work you need to essentially micro marines one by one so that they stay spread out enough to avoid getting massacred by storms and reavers, and to micro tanks so that they take out templars and reavers in priority.
While, obviously, macroing or of a trillion barracks perfectly.
Mm absolutely melt zealots and goons, so there is very little protoss can actually do. What is so striking in that game v err rsus Paralyze is that pluto won off two bases, because this army comp with that level of micro doesn’t even need a strong economic engine.
ahh, i thought you meant the TvZ of skipping turrets… yeah TvP bio would be much harder to execute, i was hoping TY returning to BW would showcase bio TvP, utilizing his sc2 baneling split/disruptor split skills to minimize scarab damage, but sadly TY didn’t stick around for too long
On September 19 2026 20:18 Zografa wrote: why does it matter if it uses map hack or not? i mean even if it does - still impressive, i remember checking those bots like 1-2 years ago and they were mediocre, this thing is on another level?
Yeah, the sad part for me is that I expect a lot of people discount it because of the maphack, while that actually isn't playing any role in how good it is. So it kind of spoiled the possibility to give it a proper reveal.
I saw a game between Pluto and Paralyze on Odyssey. This is like their 6th or 7th game already so Paralyze fully knew what kind of "player" he was dealing with.
Once again the bot dismantled him with MM and Tanks on 2 bases. I think the map hack does play a role though. The MM Tanks army was CONSTANTLY moving around the map at a level I dont even think human team melee is possible. Protoss couldn't find any good angle to fight. Surely knowing the movement of your opposition's army all the time plays a role in the bot's army movement?
EPT
