
Larian Studios
An ancient evil has returned to Baldur's Gate, intent on devouring it from the inside out. The fate of Faerun lies in your hands. Alone, you may resist. But together, you can overcome. Gather your party and return to the Forgotten Realms in a tale of fellowship and betrayal, sacrifice and survival, and the lure of absolute power. Mysterious abilities are awakening inside you, drawn from a mind flayer parasite planted in your brain. Resist, and turn darkness against itself. Or embrace corruption, and become ultimate evil.
Atlas Score
Benchmark results
12
Published presets
6
Published benchmarks
12
Tested devices
5
Performance status
Benchmark results available
Atlas notes
Baldur's Gate 3 has 12 published benchmarks. Compare the graphics profile, resolution, power mode and source for each result. Results from different test conditions are not a single performance rating for the game.
Community rating
Community ratings measure player opinion separately from the PC Metacritic score displayed as Atlas Score, with one live vote per account.
No ratings · 0 votes
Sign in and choose a star to rate this game. One live rating per account is used for the all-time score.
The monthly Top 5 counts one rating per account, game and month.
Quick facts
Recommended settings
GamersNexus tested Baldur's Gate 3 on the MSI Claw A1M Core Ultra 7 155H in its stock Super Battery profile on battery (DC PL1 20W / PL2 20W). The benchmark used 1280 × 720, DirectX 11 and the Low in-game preset, with AMD FSR 2.2 Balanced, while framerate caps and variable-refresh features were disabled and tearing was allowed. The named global preset is directly selectable; GamersNexus did not separately expand every individual sub-setting. The published chart reports 29.2 FPS average, a 9.6 FPS 1% low and a 5.0 FPS 0.1% low. VRAM allocation, exact driver and OS builds, test route, capture tool, capture duration and run count were not published and remain unspecified.
Best for
portable play where lower power draw and longer battery life matter most
Resolution
1280 × 720
TDP
20W
Average
29.2 FPS
1% Low
9.6 FPS
GamersNexus tested Baldur's Gate 3 on the MSI Claw A1M Core Ultra 7 155H in its stock Balanced profile on battery (DC PL1 30W / PL2 30W). The benchmark used 1280 × 720, DirectX 11 and the Low in-game preset, with AMD FSR 2.2 Balanced, while framerate caps and variable-refresh features were disabled and tearing was allowed. The named global preset is directly selectable; GamersNexus did not separately expand every individual sub-setting. The published chart reports 34.1 FPS average, a 9.2 FPS 1% low and a 3.8 FPS 0.1% low. VRAM allocation, exact driver and OS builds, test route, capture tool, capture duration and run count were not published and remain unspecified.
Best for
players who want a stable middle ground between image quality and smoothness
Resolution
1280 × 720
TDP
30W
Average
34.1 FPS
GamersNexus tested Baldur's Gate 3 on the MSI Claw A1M Core Ultra 7 155H in its stock Performance (Extreme Performance) profile on battery (DC PL1 35W / PL2 35W). The benchmark used 1280 × 720, DirectX 11 and the Low in-game preset, with AMD FSR 2.2 Balanced, while framerate caps and variable-refresh features were disabled and tearing was allowed. The named global preset is directly selectable; GamersNexus did not separately expand every individual sub-setting. The published chart reports 34.4 FPS average, a 10.0 FPS 1% low and a 4.1 FPS 0.1% low. VRAM allocation, exact driver and OS builds, test route, capture tool, capture duration and run count were not published and remain unspecified.
Best for
players who prioritize responsiveness and the highest practical frame rate
Resolution
1280 × 720
TDP
35W
Average
34.4 FPS
GamersNexus tested Baldur's Gate 3 on the ASUS ROG Ally Z1 Extreme in its stock Silent profile on battery (Stock Silent 10W). The benchmark used 1280 × 720, DirectX 11 and the Low in-game preset, with AMD FSR 2.2 Balanced, while framerate caps and variable-refresh features were disabled and tearing was allowed. The named global preset is directly selectable; GamersNexus did not separately expand every individual sub-setting. The published chart reports 17.3 FPS average, a 9.0 FPS 1% low and a 7.6 FPS 0.1% low. VRAM allocation, exact driver and OS builds, test route, capture tool, capture duration and run count were not published and remain unspecified.
Best for
portable play where lower power draw and longer battery life matter most
Resolution
1280 × 720
TDP
10W
Average
17.3 FPS
GamersNexus tested Baldur's Gate 3 on the ASUS ROG Ally Z1 Extreme in its stock Performance profile on battery (Stock Performance 15W). The benchmark used 1280 × 720, DirectX 11 and the Low in-game preset, with AMD FSR 2.2 Balanced, while framerate caps and variable-refresh features were disabled and tearing was allowed. The named global preset is directly selectable; GamersNexus did not separately expand every individual sub-setting. The published chart reports 30.3 FPS average, a 19.3 FPS 1% low and a 16.7 FPS 0.1% low. VRAM allocation, exact driver and OS builds, test route, capture tool, capture duration and run count were not published and remain unspecified.
Best for
players who want a stable middle ground between image quality and smoothness
Resolution
1280 × 720
TDP
15W
Average
30.3 FPS
GamersNexus tested Baldur's Gate 3 on the ASUS ROG Ally Z1 Extreme in its stock Turbo profile on battery (Stock Turbo 25W). The benchmark used 1280 × 720, DirectX 11 and the Low in-game preset, with AMD FSR 2.2 Balanced, while framerate caps and variable-refresh features were disabled and tearing was allowed. The named global preset is directly selectable; GamersNexus did not separately expand every individual sub-setting. The published chart reports 38.7 FPS average, a 24.6 FPS 1% low and a 20.0 FPS 0.1% low. VRAM allocation, exact driver and OS builds, test route, capture tool, capture duration and run count were not published and remain unspecified.
Best for
players who prioritize responsiveness and the highest practical frame rate
Resolution
1280 × 720
TDP
25W
Average
38.7 FPS
Performance data
Average FPS
23.9
See the test notes for reported graphics settings.
Test date not reported · Added to Atlas Sep 13, 2026
Test route: Crowded city street workload
Historical GamersNexus Steam Deck OLED test batch. The video was originally published February 23, 2024 and the written review followed April 4, 2024; the exact hardware test date was not published. Tested on the Steam Deck OLED 1TB on battery in a crowded city street at 1280 × 720 using DirectX 11, the Low preset and FSR 2.2 Balanced. All framerate caps were disabled and tearing was allowed. The matching chart reports 23.9 FPS average, 7.8 FPS 1% low and 5.9 FPS 0.1% low. The Steam Deck used its stock power management; the source lists a 4–15W APU range but does not report a fixed per-test wattage. Exact per-option settings, VRAM allocation, SteamOS build, capture tool, capture duration and run count remain unavailable. This is an older, separate test batch and not a September 2026 retest.
Source checked Sep 13, 2026. Checking a source is not a new performance test.
Average FPS
25.4
High Preset AF TAA.
Test date not reported · Added to Atlas Sep 9, 2026
SETTINGS: High Preset AF TAA. Historical source review published 2025-02-04. ZOTAC GAMING ZONE, ZGC-G1A1W-01; Ryzen 7 8840U, Radeon 780M, 16GB RAM, 4GB reserved for the iGPU. The source specifies Performance mode at 28W with automatic fan control; hardware specifications list a 30W short boost. Evidence: the review's game benchmark table, matching game row and graphics column, with the setting tooltip checked and table visually verified. Published average 25.4 FPS at 1920 × 1080. AF/AA labels preserve the source's notation; unspecified values are not inferred. The named preset and listed overrides are documented, but a full per-option menu, frame-generation state, cap, exact test date, capture duration, run count and driver/OS builds are unavailable. Upscaling is unspecified except where the setting explicitly says FSR off. Power connection is not identified per result. Lows and battery life remain blank. Source checking is not a new performance test.
Average FPS
30.2
Medium Preset AF TAA.
Test date not reported · Added to Atlas Sep 9, 2026
SETTINGS: Medium Preset AF TAA. Historical source review published 2025-02-04. ZOTAC GAMING ZONE, ZGC-G1A1W-01; Ryzen 7 8840U, Radeon 780M, 16GB RAM, 4GB reserved for the iGPU. The source specifies Performance mode at 28W with automatic fan control; hardware specifications list a 30W short boost. Evidence: the review's game benchmark table, matching game row and graphics column, with the setting tooltip checked and table visually verified. Published average 30.2 FPS at 1920 × 1080. AF/AA labels preserve the source's notation; unspecified values are not inferred. The named preset and listed overrides are documented, but a full per-option menu, frame-generation state, cap, exact test date, capture duration, run count and driver/OS builds are unavailable. Upscaling is unspecified except where the setting explicitly says FSR off. Power connection is not identified per result. Lows and battery life remain blank. Source checking is not a new performance test.
Average FPS
24.5
Ultra Preset AF TAA.
Test date not reported · Added to Atlas Sep 9, 2026
SETTINGS: Ultra Preset AF TAA. Historical source review published 2025-02-04. ZOTAC GAMING ZONE, ZGC-G1A1W-01; Ryzen 7 8840U, Radeon 780M, 16GB RAM, 4GB reserved for the iGPU. The source specifies Performance mode at 28W with automatic fan control; hardware specifications list a 30W short boost. Evidence: the review's game benchmark table, matching game row and graphics column, with the setting tooltip checked and table visually verified. Published average 24.5 FPS at 1920 × 1080. AF/AA labels preserve the source's notation; unspecified values are not inferred. The named preset and listed overrides are documented, but a full per-option menu, frame-generation state, cap, exact test date, capture duration, run count and driver/OS builds are unavailable. Upscaling is unspecified except where the setting explicitly says FSR off. Power connection is not identified per result. Lows and battery life remain blank. Source checking is not a new performance test.
Average FPS
30.2
See the test notes for reported graphics settings.
Test date not reported · Added to Atlas Sep 8, 2026
Historical comparison: video published 2024-04-14, article dated 2024-05-02. Steam Deck OLED; not LCD. Settings: DirectX 11; Low preset; FSR 2.2 Balanced. Frame caps and VRR disabled; tearing allowed. Manufacturer power profile on battery. Evidence: the matching 720p game chart in the linked review, visually checked against the device/profile row. Average 30.2 FPS, 1% low 10.2 FPS, 0.1% low 8.6 FPS. These lows describe the captured workload, not a guaranteed gameplay floor. Numeric TDP, VRAM allocation, full per-option menu, exact test date, software versions, run count and capture duration are not supplied. This standalone external benchmark is not a complete custom preset or a September 2026 retest.
Average FPS
38.3
See the test notes for reported graphics settings.
Test date not reported · Added to Atlas Sep 8, 2026
Historical comparison: video published 2024-04-14, article dated 2024-05-02. Lenovo Legion Go with Ryzen Z1 Extreme; not Z1 or Go S. Settings: DirectX 11; Low preset; FSR 2.2 Balanced. Frame caps and VRR disabled; tearing allowed. Manufacturer power profile on battery. Evidence: the matching 720p game chart in the linked review, visually checked against the device/profile row. Average 38.3 FPS, 1% low 23.1 FPS, 0.1% low 19.1 FPS. These lows describe the captured workload, not a guaranteed gameplay floor. Numeric TDP, VRAM allocation, full per-option menu, exact test date, software versions, run count and capture duration are not supplied. This standalone external benchmark is not a complete custom preset or a September 2026 retest.
1% Low
9.2 FPS
1% Low
10 FPS
1% Low
9 FPS
1% Low
19.3 FPS
1% Low
24.6 FPS
Source checked Sep 9, 2026. Checking a source is not a new performance test.
Source checked Sep 9, 2026. Checking a source is not a new performance test.
Source checked Sep 9, 2026. Checking a source is not a new performance test.
Source checked Sep 8, 2026. Checking a source is not a new performance test.
Source checked Sep 8, 2026. Checking a source is not a new performance test.