Noometry

Agentic & Tool Use benchmark

BALROG leaderboard

As of October 2026, GPT-6 Astra has the highest published BALROG score on Noometry at 68.3%, out of 35 models with results.

Last verified

About BALROG

Long-horizon game environments, from BabyAI to NetHack, that test planning and exploration.

Category
Agentic & Tool Use
Introduced
2024
Format
Games
Unit
Percent (random guessing ≈ 0%)
Official site
balrogai.com

Top 15 models

  1. GPT-6 Astra 68.3%
  2. Claude Opus 5 63.4%
  3. GPT-5.6 Sol 60%
  4. Gemini 3 Pro 58.1%
  5. Gemini 3.1 Pro Preview 57%
  6. GPT-5.6 Terra 53.2%
  7. Gemini 3 Flash Preview 48.1%
  8. GPT-5.6 Luna 45.6%
  9. Grok 4 43.6%
  10. Claude Opus 4.5 43.5%
  11. Gemini 2.5 Pro 43.3%
  12. DeepSeek-R1 34.9%
  13. Gemini 2.5 Flash 33.5%
  14. GPT-5 32.8%
  15. Claude 3.5 Sonnet 32.6%

Sponsored placements are available on pages like this one. Advertise on Noometry

All results

#ModelProviderScoreSettingSourceDate
1GPT-6 Astra OpenAI68.3%maxEpoch AI
2Claude Opus 5 Anthropic63.4%maxEpoch AI
3GPT-5.6 Sol OpenAI60%maxEpoch AI
4Gemini 3 Pro Google58.1%Epoch AI
5Gemini 3.1 Pro Preview Google57%Epoch AI
6GPT-5.6 Terra OpenAI53.2%maxEpoch AI
7Gemini 3 Flash Preview Google48.1%Epoch AI
8GPT-5.6 Luna OpenAI45.6%maxEpoch AI
9Grok 4 xAI43.6%Epoch AI
10Claude Opus 4.5 Anthropic43.5%Epoch AI
11Gemini 2.5 Pro Google43.3%Epoch AI
12DeepSeek-R1 DeepSeek34.9%Epoch AI
13Gemini 2.5 Flash Google33.5%Epoch AI
14GPT-5 OpenAI32.8%minimalEpoch AI
15Claude 3.5 Sonnet Anthropic32.6%Epoch AI
16GPT-4o OpenAI32.3%Epoch AI
17Claude Haiku 4.5 Anthropic31.2%Epoch AI
18Grok 3 xAI29.5%Epoch AI
19Reka Flash 3 Reka29.2%Epoch AI
20Llama 3.1-70B Meta27.9%Epoch AI
21Llama 3.2 90B Meta27.3%Epoch AI
22Llama-3.3-70B-Instruct Meta23%Epoch AI
23Gemini 1.5 Pro (May 2024) Google21%Epoch AI
24DeepSeek-R1-Distill-Qwen-32B DeepSeek19.5%Epoch AI
25Claude 3.5 Haiku Anthropic19.3%Epoch AI
26Mistral Nemo Mistral AI17.6%Epoch AI
27GPT-4o mini OpenAI17.4%Epoch AI
28Llama 3.2 11B Meta16.8%Epoch AI
29Qwen2.5 72B Instruct Alibaba (Qwen)16.2%Epoch AI
30Llama 3.1-8B Meta15.1%Epoch AI
31Gemini 1.5 Flash (May 2024) Google14.6%Epoch AI
32Phi-4 Microsoft11.6%Epoch AI
33Llama 3.2 3B Meta10.1%Epoch AI
34Qwen2.5 7B Instruct Alibaba (Qwen)7.8%Epoch AI
35Llama 3.2 1B Meta6.6%Epoch AI

Compare the leaders

Other agentic & tool use benchmarks

Frequently asked questions

What does BALROG measure?

Long-horizon game environments, from BabyAI to NetHack, that test planning and exploration.

Which model has the highest BALROG score?

As of October 2026, GPT-6 Astra has the highest published BALROG score on Noometry at 68.3%, out of 35 models with results.

What is the best open-weight model on BALROG?

Reka Flash 3 has the highest BALROG accuracy among open-weight models at 29.2%, ranking 19 of 35 overall.