Qwen3.5-35B-A3B
A large language model with a MoE architecture from Alibaba's Qwen series, with 35B total parameters and approximately 3B activated parameters, suited to scenarios such as terminal tasks.
Timeline (last 90 days)
On the Qwen3.5-35B-A3B backbone model, ClueFlow achieves 60.0% accuracy on the BrowseComp benchmark.
On Terminal-Bench 2.0, Qwen3.5-35B-A3B scores 37.1% with the previous method and 48.3% with EditAct.
Using QLoRA on A100 and RTX PRO 6000, training speed for Qwen3.5-35B-A3B can be improved by up to 4.1×, and for Qwen3-30B-A3B by up to 3.3×.
All Facts (3)
On the Qwen3.5-35B-A3B backbone model, ClueFlow achieves 60.0% accuracy on the BrowseComp benchmark.
50%UnverifiedOn Terminal-Bench 2.0, Qwen3.5-35B-A3B scores 37.1% with the previous method and 48.3% with EditAct.
50%UnverifiedUsing QLoRA on A100 and RTX PRO 6000, training speed for Qwen3.5-35B-A3B can be improved by up to 4.1×, and for Qwen3-30B-A3B by up to 3.3×.
50%