Who leads the Atari 2600 benchmark?

Go-Explore currently leads Atari 2600 with a score of 40000 on human-normalized-score.

What is the state-of-the-art score on Atari 2600?

The state-of-the-art result on Atari 2600 is 40000 (human-normalized-score), achieved by Go-Explore as of 2025.

How many models are tracked on Atari 2600?

Codesota tracks 12 models on Atari 2600.

When was the Atari 2600 leaderboard last updated?

The Atari 2600 leaderboard on Codesota includes results through 2025.

Codesota · Reinforcement Learning · Atari Games · Atari 2600Tasks/Reinforcement Learning/Atari Games

Atari Games · benchmark dataset · 2013 · N/A

Arcade Learning Environment (Atari 2600).

Name: Arcade Learning Environment (Atari 2600) Benchmark Results
Creator: Codesota
Published: 2025-01-01
License: https://creativecommons.org/licenses/by/4.0/

Suite of 57 Atari 2600 games. Standard benchmark for deep reinforcement learning agents.

Paper ↗Submit a result ↵

§ 01 · Leaderboard

Best published scores.

12 results indexed across 1 metric. Shaded row marks current SOTA; ties broken by submission date.

Primary: human-normalized-score · higher is better

human-normalized-score· primary

12 rows

#	Model	Org	Submitted	Paper / code	human-normalized-score
01	Go-ExploreOpen	Uber AI	Dec 2025	nature-paper	40000
02	LBCOpen	Tsinghua University / Baidu	—	source	10078
03	Agent57Open	DeepMind	Dec 2025	deepmind-research	4731.30
04	MEMEOpen	Google DeepMind	—	source	4087
05	Disco57Open	Google DeepMind	—	source	1386
06	BBOS-1Open	—	Dec 2025	—	1100
07	GDI-H3Open	Research	Dec 2025	—	950
08	DreamerV3Open	Google DeepMind	Dec 2025	arxiv-paper	840
09	MuZeroOpen	DeepMind	Dec 2025	nature-paper	731
10	Rainbow DQNOpen	DeepMind	Dec 2025	aaai-paper	231
11	Human Professional	Biology	Dec 2025	baseline	100
12	DQN (Human-level)Open	DeepMind	Dec 2025	nature-paper	79

Fig 2 · Rows sorted by score within each metric. Shaded row marks SOTA. Dates reflect model or paper release where available, otherwise the date Codesota accessed the source.

§ 03 · Progress

1 steps
of state of the art.

Each row below marks a model that broke the previous record on human-normalized-score. Intermediate submissions are kept in the leaderboard above; only SOTA-setting entries are re-listed here.

Higher scores win. Each subsequent entry improved upon the previous best.

SOTA line · human-normalized-score

Dec 18, 2025Go-ExploreUber AI40000

Fig 3 · SOTA-setting models only. 1 entries span Dec 2025 → Dec 2025.

§ 06 · Contribute

Have a score that beats
this table?

Submit a checkpoint and a reproduction script. We will run it, publish the score, and — if it takes the top — annotate the step on the progress chart with your name.

Submit a result ↵Read submission guide

What a submission needs

01A public checkpoint or API endpoint
02A reproduction script with frozen commit + seed
03Declared evaluation environment (Python, deps)
04One row per metric declared by this dataset
05A contact so we can follow up on discrepancies

Arcade Learning Environment (Atari 2600).

Best published scores.

1 stepsof state of the art.

Have a score that beatsthis table?

1 steps
of state of the art.

Have a score that beats
this table?