I Tested 14 Poker Bots. Here's My Tier List.

Eighteen months. Fourteen bots. Six criteria I actually care about. Below is where each one landed — with stars, scores, and a line on why. I'm not selling you any of these; I run the testbench, I write what I see.

Honestly? After eighteen months of testing the bot landscape I can tell you the gap between S-tier and A-tier is bigger than between A and D. Most bots are roughly equally mediocre — they all kind of play tight-passive, they all fold to 3-bets too much, they all panic against multi-way boards. What separates the top of this table from the bottom isn't a magical poker engine. It's whether the thing survives detection past month two and whether the people behind it answer their email.

I'm publishing this as a tier list because every other "best poker bot 2026" page I read in research was either a paid placement or a thinly-veiled affiliate funnel. Mine is just me, the bench, and a spreadsheet. Take the scores with whatever salt feels appropriate — but at least the salt is honest.

The Tier List

Five tiers (S/A/B/C/D), six criteria, weighted into a /10. Names are anonymised — naming real products in a public tier list invites the wrong kind of attention and frankly the product mix shifts every quarter anyway. What matters is the shape of the field, not which logo's on top this month.

The S-Tier

Only one bot earned S this round, and I went back and forth on it for a week before publishing. What put Bot Alpha there isn't EV — actually Bot Beta out-earns it by maybe 0.6bb/100 in the same lineup. What put it there is that it ran for fourteen months on two of the harder app-clubs I have access to without a single account flag. Fourteen months. Most of this category gets caught inside ninety days.

The other thing I'll say about S: their support actually wrote back the same day. Twice I asked weird questions ("what happens if the table layout shifts mid-hand") and got a real answer, not a templated one. That sounds soft, but it's the difference between a $4K install you can live with and a $4K install that ghosts you in week three.

The A-Tier

Beta and Gamma are the workhorses. Beta is the highest-EV bot I tested — its postflop play against regs is genuinely sharp, not the usual c-bet-and-pray. Gamma is a half-step behind on EV but it has the cleanest install flow of anything in the field; if I were standing up a club from scratch with no technical staff, Gamma is what I'd pick. Both lose points on detection — they're good, not great, and I had one Gamma seat get flagged at the four-month mark on a tight room.

The B and below

B is the "fine, if you know what you're walking into" tier. Delta and Epsilon both work, both ship updates, both will make money in soft pools. They will not survive a serious detection upgrade and their support is hit-or-miss. Below that — Zeta and Eta — I genuinely cannot recommend. Zeta has good marketing and bad code. Eta crashed three times during my test window and the operator's last update notes are from late 2024.

I rated detection-survival 35% of the total score because that's where most of these things actually die. EV is only 25% — there's no point shipping a 4bb/100 bot that gets banned in week three.

If you want the weighting math itself, I broke it out on the methodology page. If you want what each criterion actually means — "detection-survival" isn't a vibe, it's measured — that's on my criteria page.