1d ago

Feng Chen from the University of Hong Kong introduces DexHoldem, a benchmark for dexterous robots playing Texas Hold’em that shows partial success on isolated subtasks but none when combining manipulation and state reading

ShadowHand on UR10e arm uses three camera views for data collection.

——0——
Original post
Yi Ma#368@YIMATWEETSOPFeng ChenFeng Chen@WINNIECHEN02

🤖Can robots play Texas Hold’em? 🤔 We introduce DexHoldem, a real-world benchmark for dexterous manipulation and embodied agents. After fine-tuning leading VLAs and pairing them with frontier agents, we found: Robots can move the chips — sometimes. Agents can read the table state — sometimes. Unfortunately, poker requires both to happen correctly in the same hand. So no, robots are not ready to take your money at the casino — yet. More below 👇

9:33 PM · May 18, 2026 View on X
QUOTE POSTCLS#442CLS@CHENGLEISI

Gonna buy a lot of these when I retire

Feng ChenFeng Chen@Winniechen02

🤖Can robots play Texas Hold’em? 🤔 We introduce DexHoldem, a real-world benchmark for dexterous manipulation and embodied agents. After fine-tuning leading VLAs and pairing them with frontier agents, we found: Robots can move the chips — sometimes. Agents can read the table state — sometimes. Unfortunately, poker requires both to happen correctly in the same hand. So no, robots are not ready to take your money at the casino — yet. More below 👇

4:33 AM · May 19, 2026 · 4.6K Views
9:54 PM · May 20, 2026 · 365 Views
Feng Chen from the University of Hong Kong introduces DexHoldem, a benchmark for dexterous robots playing Texas Hold’em that shows partial success on isolated subtasks but none when combining manipulation and state reading · Digg