open-agent-leaderboard
/

smolagent

agent-evaluation

Model card Files Files and versions

Smolagent

This is a tracking repo for Smolagent, used by the Open Agent Leaderboard to report evaluation results on HuggingFace.

Hugging Face's lightweight agent framework using code-based actions rather than JSON tool calls.

Framework: smolagents
Leaderboard: Open Agent Leaderboard
Paper: arXiv:2602.22953

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for open-agent-leaderboard/smolagent

General Agent Evaluation

Paper • 2602.22953 • Published Feb 26 • 12

Evaluation results

Overall
on open-agent-leaderboard/results View evaluation results source leaderboard

model: Claude Opus 4.5
0.66 ^*
on open-agent-leaderboard/results View evaluation results source leaderboard

model: DeepSeek V3.2
0.41 ^*
Appworld on open-agent-leaderboard/results View evaluation results source leaderboard

0.7 ^*
Browsecomp Plus on open-agent-leaderboard/results View evaluation results source leaderboard

0.61 ^*
Swebench on open-agent-leaderboard/results View evaluation results source leaderboard

0.65 ^*
Taubench Airline on open-agent-leaderboard/results View evaluation results source leaderboard

0.72 ^*
Taubench Retail on open-agent-leaderboard/results View evaluation results source leaderboard

0.78 ^*
Taubench Telecom on open-agent-leaderboard/results View evaluation results source leaderboard

0.58 ^*