This is very cool, one piece of feedback: watching the table as the AI plays while seeing the reasoning is difficult as they're on other sides of the screen. It could be nice to have the reasoning show up next to the players as they make their moves.
I just saw GPT 5.2 do something absurd. It has a crazy amount of money ($26k) but folded with a 4-pair before the flop. That's insanely conservative, when it would have cost just $20 to see the flop. But even worse, on the very next hand it decided to place $20 down with a 5 and 4 of different suits.
In fact, all of them love folding before the flop. Most of the hands I'm seeing go like - $10 (small blind), $20 (big blind), fold, $70 bet, everyone folds. The site says "won $100", but in most of these cases that one LLM is picking up the blinds alone - $30. Chump change.
This is illuminating, but not a resource for learning poker.
Modern poker (which tbf not sure if these LLMs are acting according to modern GTO or not) is highly dependent on position. Things change a lot too when/if you are in SB/BB.
Idea: can the agents make faces?
1. Programmatically--agents see each other's faces, and they can make their own. They can choose to ignore, but at least make that an input to the decision making.
2. Display them in UI--I just want to see their faces instead next to their model code names :)
OK so you know how it goes in poker and I should probably read the literature ...
How much of a session is based on "reading players" vs "playing the odds"?
What I am getting at, is how different is poker than say roulette or blackjack? My initial thoughts are that poker such as TX hold 'em is not a game offered in a casino, so it must be mostly indeterminate. I imagine that the casino versions of poker are not TXHT.
By contrast, roulette is simply a game where the casino wins eventually with a fixed profit (thanks to 0 and a possible 00). That is all well documented.
I have only ever visited a casino once, 25 years ago, Plymouth, Devon as it turns out and I was advised to only take £50 in readies and bail out when it was gone. I came out £90 up, which was nice and my "advisor" came out £95 up (eventually, after being £200 down at one point). Sadly my "advisor" ended up bankrupt a year later.
So, how do you play a LLM? I would imagine that conversation is not allowed ...
Just an fyi they make apps like https://apps.apple.com/app/id1530767783 to train on betting based on expected value.Not the whole picture but trains the math side.
They offer (real) poker at some casinos. It's standard NLHE usually 100-200bb max buyin, sometimes match the stack etc.
Most common game spread is 9-handed $200 max $1/$2 NLHE. It's exactly like the game on the link, except more players and lower stakes.
In the game, you try to win the money of the other 8 players, not of the casino. The casino takes a rake each hand, and a player with a large enough edge can overcome it. The edge might be you're excellent, or it might be they're terrible (or drunk). But the house gets paid to deal each hand.
In the long term, poker outcomes are determined by skill. In the short term, they're luck. In the medium term, both. Most people never reach the long term, it's a lot of hands.
There's also table games, similar to blackjack, that they call "three card poker" etc. These can't be beat, they favor the house. Standard table game, with a poker flavor. I've never played one of these.
> How much of a session is based on "reading players" vs "playing the odds"?
Several good answers here, but I will add my own take.
“Playing the odds” is basically playing good, fundamental poker. This is a baseline that most players will use when sitting down at a table of unknown players. This is often called a “balanced” strategy (note some people erroneously call this “GTO strategy, assuming the balance, which is not actually what GTO is).
“Reading players” is a thing, but it can be broken up into (at least) two categories: 1) physical tells, and 2) player habits.
Physical tells is not a big thing. Some people give off a lot of tells, but some folks are also decent (not good) at giving reverse tells. Honestly, you can be a wildly successful poker player while knowing nothing about physical tells.
(Side note: One of the most reliable tells is bet timing tells. This can sometimes even be a tell online, especially when people are taking shots at higher stakes or are deep in a tournament. It can also be faked, but some folks are super reliable with timing tells, and they don’t realize it.)
The second kind of tell is player tendencies — things like when players play too many hands or when they play too few (e.g., fold too often).
One very reliable way for good players to smooth out their earnings curve is figure out which players fold too often and in which spots. Once they’ve figured that out, they try to set up that spot and basically print low-risk (sometimes even no-risk) chips.
Taking advantage of these tendencies is called an “exploitative” strategy (as opposed to the “balanced” strategy mentioned above).
Really good players can take rec players on a journey through a series of emotions (and accompanying predictable gameplay) such that the good player can read the rec player like a book. The odds tip heavily in favor of the good player at this point.
Pro player and strong amateur players are so far ahead of recs in ways that the recs don’t even realize.
>How much of a session is based on "reading players" vs "playing the odds"?
I think the key is you need to watch for a person's play style.
There's a two axis system: tight/agressive and passive/active.
An active player sees more flops, and an aggressive player will call and raise more than a tight player.
So a tight, aggressive player sees few flows but bets strongly when they have a good hand -- this is considered "good" strategy.
Others might play a "tight-passive" strategy -- they'll play few hands but fold easily. They won't lose large amounts of money but they'll slowly bleed chips.
A loose, aggeessive player is the type you want at the table -- they're making a lot of bets, and often bluffing, and you can sit and wait to catch them.
Now, this is "reading" someone, but it's not the Rounders style "oh he just ate an oreo so he's bluffing" level reading of a player that movies
For context, I'm an OK player. I can make a few hundred playing 1/3 per session -- I'm not in Vegas so I can't move to the next tier without sinking a lot of money on a flight and hotel.
If your goal is a bit of beer money, it can be a fun hobby, but I wouldn't go into it expecting it to become a full time career.
At low levels, playing is ABC simple and mostly about following basic strategy for starting hands and pot adds for chasing. Don’t get fancy and keep your temperament steady and you’ll win.
To a slight degree, you can do better with reading players and identifying them in broad ways (wild, conservative, confused, etc.) but don’t let that allow you to get fancy. Stick to the basic fundamental strategy for hands, position, and pot odds to crush lower level games.
Hold'em is offered in casinos routinely, I'm not sure where else one even goes to play it aside from private games, but it is not against the casino. It's against other players, and the casino takes a percentage of the pot.
Others may differ and I am biased because 99% of my play has been online, but I'd say it's almost entirely playing the odds. Or at least, the popular romantic conception of looking for tells or whatever, is, I would expect, a really minimal edge compared to simply playing better.
You do learn the other players' tendencies and adapt accordingly, and table selection is very important, so in that sense it is very much about reading players.
A large part of my play was heads up where it's very much about understanding the other player's play as deeply as possible, and so if I wanted to be technically accurate about reading players vs playing the odds, I'd say both are very important. But if I'm answering someone who has the popular conception of what those phrases mean, I think saying "it's about playing the odds" would give them the more accurate picture.
You really want to be good at playing the odds, and you don't want to stray too far from fundamentally good play. If someone is learning how to play and I'm advising them, I'm teaching them all about playing the odds, and trying to get them to read players less. Only once they have a solid fundamental understanding of the odds would I teach them how to adjust.
Around here (Melbourne) the other place is in pubs - there are organised poker tournaments. They can't legally charge you an entry fee, but they can give you a lot of extra chips if you buy a meal at the pub. Some modest prizes if you win.
They're kind of a ridiculous format - you typically start with about 20 BB but the blinds go up pretty quickly so you don't see a lot of post-flop play.
> My initial thoughts are that poker such as TX hold 'em is not a game offered in a casino
Why not? Because you think it's a game where the casino can lose?
If so it's not an issue, as casinos that provide poker take "fees" from the stakes. Like how stock exchanges work: there are people making or losing money from stock market, but exchanges are always making profit.
Around where I live, about half the casinos offer poker and about half don't. Poker can be a pretty high cost to a casino. Compared to something like slot machines, it's financially mostly downside: You need a lot more physical space per player. You need more staff. In order for players to come and actually enjoy it, the poker room needs to be located in a relatively quiet corner, ideally enclosed so the buzz of the rest of the casino can't be heard, which is also expensive. And the game is slow, and rakes happen once per hand, so you're making money pretty slowly. And that's just for cash games. Tournaments are worse. If I had to guess, tournaments probably lose money for the casino, and they only exist to get players in so they play at the cash tables. Probably many other things that I'm forgetting because I don't run a casino.
I just watched for 5 min and no they don't play very well. Deepseek squeezed with K4o against CO open and BTN call with full stacks. Grok 3b AI with 25bb in the button with Q4s. Those are very far from optimal play which is well known since solvers. I wonder how they've been trained.
there plenty of published preflop charts and GTO ranges
in fact, a fun project would be take a non-reasoning model, play on a lesser known game format, and see if it learns an "a ha" moment or explicitly simulate moves ahead
On Pokerstars it is the right move because you are going to get beat by someone going all in with 72o or something
but seriously at lower stakes there is just no respect for the art its just a shock and awe strategy: throw shit up, break the game and use that demoralization to bully others.
Thanks for building and sharing, looks cool and is very entertaining.
I had similar idea for people to code poker playing bots and enter tournaments versus each other, this was pre-llm, however.
It would be fun if you hosted a 'tournament' every month and had each of the latest releases from the major models participate and see who comes out on top.
Or perhaps do open it up to others to enter and participate versus each other - where they can choose the model they want to build with and also enter custom prompt instructions to mold the play as they wish.
Good question! The player rooms have a rate limit per day. And as for the main table, it's actually a replay of hands I recorded the LLMs playing against each other over an extended time which eventually loops.
Looks like this was cleverly designed to prevent costs blowing up. There's one game shared for everyone on the main page, and up to 100 private games per day.
I also started working on a similar project, but I think that LLM should know and be able to keep internal statistics about players. In poker, the best hand does not always win. Often, you can win by using emotions/words. LLM should be given the ability to communicate, mislead, etc.
Given online is now bot-riddled, I half-finished something similar a while back, where the game was adopting and 'coaching' (a <500 character prompt was allowed every time the dealer chip passed, outside of play) an LLM player, as a kind of gambling-on-how-good-at-prompting-you-are game. Feature request! The rake could pay for the tokens, at least.
Do the players (LLMs) have memory of how prior hands were played by their opponents, or know their VPIP and PFR percentages? Or is each hand stateless?
> Not really. Only as far as their table image mattered--in this case, zero.
Right; there's feedback to it. When humans play poker, they do so with common knowledge of the fact that humans have object permanence and can recognize and remember their opponents. The same thing that motivates "profiling" a villain, motivates attempting to project a table image, which in turn motivates being aware of the table image one is projecting.
Thank you, I'll try to grab a table when it resets :) ! I've been getting into poker (always wanted to) since I found a lecture series from John Hopkins, and severely disappointed by my options to play online in NY (real or fake money). I just want to get reps in
I'm not an expert, but as I understand it there are existing solvers for poker/holdem? Perhaps one of the players could be a traditional solver to see how the LLMs fare against those?
This also wouldn't even be a close contest, I think Pluribus demonstrated a solid win rate against professional players in a test.
As I was developing this project, a main thought came to mind as to the comparison between cost and performance between a "purpose" built AI such as Pluribus versus a general LLM model. I think Pluribus training costs ~$144 in cloud computing credits.
To expand on this - an LLM will try to play (and reason) like a person would, while a solver simply crunches the possibility space for the mathematically optimal move.
It’s similar to how an LLM can sometimes play chess on a reasonably high (but not world-class) level, while Stockfish (the chess solver) can easily crush even the best human player in the world.
GTO (“game theory optimal”) poker solvers are based around a decision tree with pre-set bet sizes (eg: check, bet small, bet large, all in), which are adjusted/optimized for stack depth and position. This simplifies the problem space: including arbitrary bet sizes would make the tree vastly larger and increase computational cost exponentially.
No, I'm not super certain, but I believe most solvers are trained to be game theory optimal (GTO), which means they assume every other player is also playing GTO. This means there is no strategy which beats them in the long run, but they may not be playing the absolute best strategy.
Not only to limit the scope of what it has to simulate, but only a certain number of bet sizes is practical for a human to implement in their strategy.
How would an LLM play like a human would? I kind of doubt that there is enough recounting of poker hands or transcription of filmed poker games in the training data to imbue a human-like decision pattern.
"Solvers" normally means algorithms which aim to produce some mathematically optimal (given certain assumptions) behaviour.
There are other poker playing programs [0] - what we called AI before large language models were a thing - which achieve superhuman performance in real time in this format. They would crush the LLMs here. I don't know what's publicly available though.
Well they can be watching all the action and thinking the whole time as the action leads up them, just like we do in poker. To me it's different, subtly perhaps.
For my implementation, I'm passing in the current hand's action history (e.g. Player 1 raises to $X preflop, Player 2 calls, Player 3 calls. Flop is A B C, Player 2 checks, etc) whenever the action is on the player.
Your idea of having it being passed in real time and having the LLM create a chain of thoughts even if action is not on them is interesting. I'd be curious to see if it would result in improved play.
Needs a four color deck, and the colors on the cards of the waiting players should not be monochrome - makes it hard to evaluate what's happening in the hand. Also, a dealer button on the table would help in visually following the action.
If you are interested in this space, you can check out NovaSolver.com
It's mostly a ChatGPT conversational interface over a classic Solver (Monte-Carlo simulation based), but that ease of use makes it very convenient for quick post-game analysis of hands.
I'm sure if you hook a Solver to a hud, it might be even simpler, but it's quite burdensome for amateurs, and it might be too close to cheating.
In fact, all of them love folding before the flop. Most of the hands I'm seeing go like - $10 (small blind), $20 (big blind), fold, $70 bet, everyone folds. The site says "won $100", but in most of these cases that one LLM is picking up the blinds alone - $30. Chump change.
This is illuminating, but not a resource for learning poker.
These LLMs are playing better than most human players I encounter (low limits).
They're kinda bad, but not as criminally bad as the humans.
How much of a session is based on "reading players" vs "playing the odds"?
What I am getting at, is how different is poker than say roulette or blackjack? My initial thoughts are that poker such as TX hold 'em is not a game offered in a casino, so it must be mostly indeterminate. I imagine that the casino versions of poker are not TXHT.
By contrast, roulette is simply a game where the casino wins eventually with a fixed profit (thanks to 0 and a possible 00). That is all well documented.
I have only ever visited a casino once, 25 years ago, Plymouth, Devon as it turns out and I was advised to only take £50 in readies and bail out when it was gone. I came out £90 up, which was nice and my "advisor" came out £95 up (eventually, after being £200 down at one point). Sadly my "advisor" ended up bankrupt a year later.
So, how do you play a LLM? I would imagine that conversation is not allowed ...
Most common game spread is 9-handed $200 max $1/$2 NLHE. It's exactly like the game on the link, except more players and lower stakes.
In the game, you try to win the money of the other 8 players, not of the casino. The casino takes a rake each hand, and a player with a large enough edge can overcome it. The edge might be you're excellent, or it might be they're terrible (or drunk). But the house gets paid to deal each hand.
In the long term, poker outcomes are determined by skill. In the short term, they're luck. In the medium term, both. Most people never reach the long term, it's a lot of hands.
There's also table games, similar to blackjack, that they call "three card poker" etc. These can't be beat, they favor the house. Standard table game, with a poker flavor. I've never played one of these.
Several good answers here, but I will add my own take.
“Playing the odds” is basically playing good, fundamental poker. This is a baseline that most players will use when sitting down at a table of unknown players. This is often called a “balanced” strategy (note some people erroneously call this “GTO strategy, assuming the balance, which is not actually what GTO is).
“Reading players” is a thing, but it can be broken up into (at least) two categories: 1) physical tells, and 2) player habits.
Physical tells is not a big thing. Some people give off a lot of tells, but some folks are also decent (not good) at giving reverse tells. Honestly, you can be a wildly successful poker player while knowing nothing about physical tells.
(Side note: One of the most reliable tells is bet timing tells. This can sometimes even be a tell online, especially when people are taking shots at higher stakes or are deep in a tournament. It can also be faked, but some folks are super reliable with timing tells, and they don’t realize it.)
The second kind of tell is player tendencies — things like when players play too many hands or when they play too few (e.g., fold too often).
One very reliable way for good players to smooth out their earnings curve is figure out which players fold too often and in which spots. Once they’ve figured that out, they try to set up that spot and basically print low-risk (sometimes even no-risk) chips.
Taking advantage of these tendencies is called an “exploitative” strategy (as opposed to the “balanced” strategy mentioned above).
Really good players can take rec players on a journey through a series of emotions (and accompanying predictable gameplay) such that the good player can read the rec player like a book. The odds tip heavily in favor of the good player at this point.
Pro player and strong amateur players are so far ahead of recs in ways that the recs don’t even realize.
I think the key is you need to watch for a person's play style.
There's a two axis system: tight/agressive and passive/active.
An active player sees more flops, and an aggressive player will call and raise more than a tight player.
So a tight, aggressive player sees few flows but bets strongly when they have a good hand -- this is considered "good" strategy.
Others might play a "tight-passive" strategy -- they'll play few hands but fold easily. They won't lose large amounts of money but they'll slowly bleed chips.
A loose, aggeessive player is the type you want at the table -- they're making a lot of bets, and often bluffing, and you can sit and wait to catch them.
Now, this is "reading" someone, but it's not the Rounders style "oh he just ate an oreo so he's bluffing" level reading of a player that movies
For context, I'm an OK player. I can make a few hundred playing 1/3 per session -- I'm not in Vegas so I can't move to the next tier without sinking a lot of money on a flight and hotel.
If your goal is a bit of beer money, it can be a fun hobby, but I wouldn't go into it expecting it to become a full time career.
At low levels, playing is ABC simple and mostly about following basic strategy for starting hands and pot adds for chasing. Don’t get fancy and keep your temperament steady and you’ll win.
To a slight degree, you can do better with reading players and identifying them in broad ways (wild, conservative, confused, etc.) but don’t let that allow you to get fancy. Stick to the basic fundamental strategy for hands, position, and pot odds to crush lower level games.
Others may differ and I am biased because 99% of my play has been online, but I'd say it's almost entirely playing the odds. Or at least, the popular romantic conception of looking for tells or whatever, is, I would expect, a really minimal edge compared to simply playing better.
You do learn the other players' tendencies and adapt accordingly, and table selection is very important, so in that sense it is very much about reading players.
A large part of my play was heads up where it's very much about understanding the other player's play as deeply as possible, and so if I wanted to be technically accurate about reading players vs playing the odds, I'd say both are very important. But if I'm answering someone who has the popular conception of what those phrases mean, I think saying "it's about playing the odds" would give them the more accurate picture.
You really want to be good at playing the odds, and you don't want to stray too far from fundamentally good play. If someone is learning how to play and I'm advising them, I'm teaching them all about playing the odds, and trying to get them to read players less. Only once they have a solid fundamental understanding of the odds would I teach them how to adjust.
They're kind of a ridiculous format - you typically start with about 20 BB but the blinds go up pretty quickly so you don't see a lot of post-flop play.
Somewhat entertaining.
Why not? Because you think it's a game where the casino can lose?
If so it's not an issue, as casinos that provide poker take "fees" from the stakes. Like how stock exchanges work: there are people making or losing money from stock market, but exchanges are always making profit.
Post-flop on the other hand is all over the place...
in fact, a fun project would be take a non-reasoning model, play on a lesser known game format, and see if it learns an "a ha" moment or explicitly simulate moves ahead
but seriously at lower stakes there is just no respect for the art its just a shock and awe strategy: throw shit up, break the game and use that demoralization to bully others.
I had similar idea for people to code poker playing bots and enter tournaments versus each other, this was pre-llm, however.
It would be fun if you hosted a 'tournament' every month and had each of the latest releases from the major models participate and see who comes out on top.
Or perhaps do open it up to others to enter and participate versus each other - where they can choose the model they want to build with and also enter custom prompt instructions to mold the play as they wish.
If you walk this path, would love to chat more.
Given online is now bot-riddled, I half-finished something similar a while back, where the game was adopting and 'coaching' (a <500 character prompt was allowed every time the dealer chip passed, outside of play) an LLM player, as a kind of gambling-on-how-good-at-prompting-you-are game. Feature request! The rake could pay for the tokens, at least.
What I'm curious about is if their innate training is enough to give them biases. Like maybe they think Grok is full of shit.
Right; there's feedback to it. When humans play poker, they do so with common knowledge of the fact that humans have object permanence and can recognize and remember their opponents. The same thing that motivates "profiling" a villain, motivates attempting to project a table image, which in turn motivates being aware of the table image one is projecting.
https://youtube.com/@jhupoker4850
https://hopkinspokercourse.com
This also wouldn't even be a close contest, I think Pluribus demonstrated a solid win rate against professional players in a test.
As I was developing this project, a main thought came to mind as to the comparison between cost and performance between a "purpose" built AI such as Pluribus versus a general LLM model. I think Pluribus training costs ~$144 in cloud computing credits.
It’s similar to how an LLM can sometimes play chess on a reasonably high (but not world-class) level, while Stockfish (the chess solver) can easily crush even the best human player in the world.
To limit the scope of what it has to simulate.
It's unlikely they're perfect, but there's very small differences in EV betting 100% vs 101.6% or whatever.
Stockfish isn't really a solver it's a neural net based engine
There are other poker playing programs [0] - what we called AI before large language models were a thing - which achieve superhuman performance in real time in this format. They would crush the LLMs here. I don't know what's publicly available though.
[0] e.g. https://en.wikipedia.org/wiki/Pluribus_(poker_bot)
Like piosolver, as an example.
The best poker-playing AI is not beatable by anyone, so yes, it would crush the LLMs.
I was interested in this idea too and made a video where some of the previous top LLMs play against each other https://www.youtube.com/watch?v=XsvcoUxGFmQ&t=2s
Or do you mean - each agent has a chance to think after every turn?
Your idea of having it being passed in real time and having the LLM create a chain of thoughts even if action is not on them is interesting. I'd be curious to see if it would result in improved play.
That is, good enough to compete amongst each other but not good enough to for one to win.
It's mostly a ChatGPT conversational interface over a classic Solver (Monte-Carlo simulation based), but that ease of use makes it very convenient for quick post-game analysis of hands.
I'm sure if you hook a Solver to a hud, it might be even simpler, but it's quite burdensome for amateurs, and it might be too close to cheating.