Qwen 3.6 27B Scores 2% on DeepSWE Benchmark

Core summary

One sentence to understand this update

The Qwen 3.6 27B model achieved a 2% score on the DeepSWE benchmark, ranking 18th out of 20, outperforming models like Haiku 4.5 and Minimax M2.7.

Impact & opportunity

What this could mean

Builders evaluating large language models can refer to such benchmark results to understand the performance of different models on specific tasks.