Browse State-of-the-Art › Atari Games

Atari Games

313 papers with code · 65 benchmarks · 6 datasets archive 2025-07-28

Playing Games

The Atari 2600 Games task (and dataset) involves training an agent to achieve high game scores.

( Image credit: Playing Atari with Deep Reinforcement Learning )

Description from the archive archive 2025-07-28.

Benchmarks archive 2025-07-28

65 leaderboard tables shown for this task, 65 with rows (a “benchmark” on this site is a table with at least one row, as on /sota), ordered by row count. “Best model” is the first row in the archive's own order at snapshot; nothing is re-ranked here and metric direction is not recorded in the archive. PwC's Trend sparklines are not in the archive, so that column is omitted. 10 shown of 65 until expanded.

DatasetBest model (first row in archive order)PaperCodeSyntologyCompare
Atari 2600 Freeway (59 rows) TRPO-hash #Exploration: A Study of Count-Based Exploration for Deep... code — Compare
Atari 2600 Breakout (58 rows) GDI-H3(200M frames) Generalized Data Distribution Iteration — — Compare
Atari 2600 Q*Bert (57 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Seaquest (57 rows) GDI-H3(200M frames) Generalized Data Distribution Iteration — — Compare
Atari 2600 Space Invaders (55 rows) GDI-H3(200M frames) Generalized Data Distribution Iteration — — Compare
Atari 2600 Venture (55 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Frostbite (53 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Gravitar (53 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Pong (52 rows) Duel noop Dueling Network Architectures for Deep Reinforcement Learning code Syntology 7 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 3 honoured, 0 violated, 2 with no contract checked; 2 where Syntology's instrument failed) · 4 unverified Compare
Atari 2600 Private Eye (52 rows) Go-Explore First return, then explore code Syntology official (archive's flag): 2 ran · 2 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified Compare
Atari 2600 Montezuma's Revenge (50 rows) Go-Explore First return, then explore code Syntology official (archive's flag): 2 ran · 2 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified Compare
Atari 2600 Alien (49 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Asterix (49 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Beam Rider (49 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Crazy Climber (49 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Amidar (48 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Enduro (48 rows) GDI-I3 GDI: Rethinking What Makes Reinforcement Learning Different From... — — Compare
Atari 2600 Battle Zone (47 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Kangaroo (47 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Ms. Pacman (47 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Demon Attack (46 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Assault (45 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Bank Heist (45 rows) MuZero (Res2 Adam) Online and Offline Reinforcement Learning by Planning with a Learned Model code — Compare
Atari 2600 Boxing (45 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Centipede (45 rows) Go-Explore First return, then explore code Syntology official (archive's flag): 2 ran · 2 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified Compare
Atari 2600 Chopper Command (45 rows) GDI-H3 GDI: Rethinking What Makes Reinforcement Learning Different From... — — Compare
Atari 2600 James Bond (45 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Krull (45 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Bowling (44 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Fishing Derby (44 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 HERO (44 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Road Runner (44 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Time Pilot (44 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Tutankham (44 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Up and Down (44 rows) GDI-I3 GDI: Rethinking What Makes Reinforcement Learning Different From... — — Compare
Atari 2600 Asteroids (43 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Double Dunk (43 rows) UCT The Arcade Learning Environment: An Evaluation Platform for General Agents code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified Compare
Atari 2600 Gopher (43 rows) GDI-I3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Ice Hockey (43 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Kung-Fu Master (43 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Name This Game (43 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Star Gunner (43 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Tennis (43 rows) GDI-I3 GDI: Rethinking What Makes Reinforcement Learning Different From... — — Compare
Atari 2600 Atlantis (42 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Robotank (42 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Video Pinball (42 rows) R2D2 Recurrent Experience Replay in Distributed Reinforcement Learning code — Compare
Atari 2600 River Raid (41 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Wizard of Wor (41 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Zaxxon (41 rows) MuZero Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model code Syntology 43 ran (of which 36 constructed an object rather than computing a result; 43 with no instrument failure: 4 honoured, 0 violated, 39 with no contract checked; 0 where Syntology's instrument failed) · 21 unverified Compare
Atari 2600 Berzerk (39 rows) Go-Explore First return, then explore code Syntology official (archive's flag): 2 ran · 2 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified Compare
Atari 2600 Pitfall! (23 rows) Go-Explore Go-Explore: a New Approach for Hard-Exploration Problems code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified Compare
Atari 2600 Skiing (23 rows) Best Learner The Arcade Learning Environment: An Evaluation Platform for General Agents code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified Compare
Atari 2600 Solaris (23 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Defender (21 rows) CGP Evolving simple programs for playing Atari games code Syntology 7 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified Compare
Atari 2600 Phoenix (21 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Yars Revenge (17 rows) Agent57 Agent57: Outperforming the Atari Human Benchmark code Syntology 8 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 0 violated, 8 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified Compare
Atari 2600 Surround (15 rows) NoisyNet-Dueling Noisy Networks for Exploration code Syntology 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified Compare
Atari games (12 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari-57 (11 rows) LBC Learnable Behavior Control: Breaking Atari Human World Records via... — — Compare
atari game (9 rows) GDI-H3 Generalized Data Distribution Iteration — — Compare
Atari 2600 Elevator Action (3 rows) Persistent AL Increasing the Action Gap: New Operators for Reinforcement Learning code — Compare
Atari 2600 Pooyan (3 rows) UCT The Arcade Learning Environment: An Evaluation Platform for General Agents code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified Compare
Atari 2600 Carnival (1 row) UCT The Arcade Learning Environment: An Evaluation Platform for General Agents code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified Compare
Atari 2600 Journey Escape (1 row) UCT The Arcade Learning Environment: An Evaluation Platform for General Agents code Syntology 2 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified Compare
Atari Pong (1 row) GATN-TL-Atari Pong Generalized Adaptive Transfer Network: Enhancing Transfer Learning... code — Compare

Syntology column: samples harvested from the paper's repositories and executed on synthesized fixtures; “ran” is not a correctness claim and does not order the table. Each line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the row's paper, the line starts with that repository's state (“official” is the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from. A dash means no Syntology record for that paper, not a recorded non-run. Each “ran” line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Libraries

Not in the archive: the export carries no per-task library table, so there is nothing to show at snapshot 2025-07-28.

Datasets archive 2025-07-28

6 datasets whose archive record lists this task, ordered by the archive's paper count.

Subtasks archive 2025-07-28

1 subtask in the archive's task tree.

Parent tasks archive 2025-07-28

Papers archive 2025-07-28

archive papers tagged: 625 · with a code link: 313 · where Syntology ran a sample: 118 (95 with a run with no instrument failure, 23 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (118 of 625 tagged: 95 with a run with no instrument failure, 23 where every run was a failure of Syntology's instrument)

Page 1 of 7: papers 1 to 100 of 625, in archive order: by repositories listed in the archive (most first), then newest first, not by stars (the archive holds no stars, so PwC's “Social” and “Latest” sorts cannot be reproduced). Papers that list no repository come after every paper that lists one.

Papers without a page here are shown as plain text. A Syntology line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the figure “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code); hover it for the repositories the samples that ran came from. The first sentence of the abstract is shown for the first 30 papers; every abstract is on its paper's page.

Syntology lines on 67 of the papers shown; no Syntology record for the others (a paper without an arXiv id cannot be joined to the graph, and absence from the graph layer is not a recorded non-run). “Ran” means the sample executed on a synthesized fixture, not that the paper's result was reproduced; each line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections