BALROG evaluates long-horizon planning, exploration, and spatial reasoning across BabyAI, Crafter, TextWorld, Baba Is AI, MiniHack, and NetHack. The headline metric averages environment completion progress.
Browse the latest scores, model modes, release dates, and parameter sizes for BALROG.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
![]() Gemini 3 Pro Standard ModeTools | 58.10 | 2026-02-01 | Unknown | Closed | |
![]() Gemini 3.1 Pro Preview Standard ModeTools | 57.00 | 2026-02-20 | Unknown | Closed | |
![]() Gemini 3.0 Flash Standard ModeTools | 48.10 | 2025-12-17 | Unknown | Closed | |
4 | Grok 4 Standard ModeTools | 43.60 | 2025-07-10 | Unknown | Closed |
5 | ![]() Opus 4.5 Standard ModeTools | 43.50 | 2025-11-25 | Unknown | Closed |
6 | ![]() Gemini 2.5 Pro Experimental 03-25 Standard ModeTools | 43.30 | 2025-03-25 | Unknown | Closed |
7 | ![]() Opus 4.5 64KTools | 43.00 | 2025-11-25 | Unknown | Closed |
8 | ![]() DeepSeek-R1 Standard ModeTools | 34.90 | 2025-01-20 | 671B | Free Commercial |
9 | ![]() Gemini 2.5 Flash Standard ModeTools | 33.50 | 2025-04-17 | Unknown | Closed |
10 | ![]() Claude 3.5 Sonnet New Standard ModeTools | 32.60 | 2024-10-22 | Unknown | Closed |
11 | ![]() Llama3.1-70B-Instruct Standard ModeTools | 27.90 | 2024-07-23 | 70B | Free Commercial |
12 | ![]() GPT-4o mini Standard ModeTools | 17.40 | 2024-07-18 | Unknown | Closed |
In the official leaderboard snapshot observed on 2026-08-12, Gemini 3 Pro Preview reached 58.1% average progress.