tussiiiii/Qwen3-4B-AgentBench-Merged
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Feb 18, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

The tussiiiii/Qwen3-4B-AgentBench-Merged model is a 4 billion parameter language model fine-tuned from Qwen/Qwen3-4B-Instruct-2507, optimized for multi-turn agent task performance. It excels at household tasks (ALFWorld) and database operations (DBBench) by learning environment observation, action selection, and tool use. This model is specifically designed to improve agentic capabilities and error recovery in complex, multi-turn interactions, leveraging a 32K context length.

Loading preview...