Valve estimates that the Steam Deck OLED delivers 3–12 hours of gameplay, depending on the content. That broad range describes the Deck OLED; it does not rank the ROG Xbox Ally X20 or Lenovo Legion Go Gen 2. A battery’s capacity alone cannot tell you how long a particular game will run.
Valve’s published Steam Deck battery figures
Valve lists a 50 Wh battery for the Steam Deck OLED and a 40 Wh battery for the Steam Deck LCD. A watt-hour (Wh) measures stored energy. Valve’s gameplay estimates vary by content, so they are not promises for a particular game or settings.
| Model | Battery capacity | Gameplay range listed by Valve |
| Steam Deck OLED | 50 Wh | 3–12 hours, depending on content |
| Steam Deck LCD | 40 Wh | 2–8 hours, depending on content |
Valve also says the OLED model has 30–50% more battery life than the LCD model. That is Valve’s broad comparison, not a game-by-game runtime result.
Why battery capacity does not decide runtime
Runtime depends on how quickly a device uses its stored energy. The game and its settings affect that power draw, so a larger battery does not by itself guarantee more hours of play. A rough runtime projection divides battery energy in watt-hours by average power use in watts; it remains a projection for the workload being measured.
That is why Steam Deck figures cannot be used to name a battery-life winner between the Deck and another handheld. The 3–12-hour OLED range is a manufacturer estimate across content, not a matched test against the ROG Xbox Ally X20 or Lenovo Legion Go Gen 2.
What a fair handheld battery test needs
A useful comparison keeps the workload and settings consistent across devices, then measures power use and frame rate over the same interval. If the battery log covers a different stretch of gameplay than the frame-rate log, an energy-per-frame calculation combines measurements from unlike periods.
Shattered.io’s October 28, 2026 tutorial labels its seven-run example, using three demo devices, as synthetic rather than measurements of the named handhelds. Its proposed procedure also gives conflicting starting-charge instructions: 100% in the prerequisites and 90% in the baseline. It describes battery logging over 20–30 minutes but a 300-second frame capture. The prose calls for a 60-second frame-time skip, while its sample manifest specifies zero. Those settings and timestamps would need to be consistent for the resulting measurements to support a fair comparison.
