Alibaba Qwen3.8 27B Tops Open Image to WebDev Models to Match 2.8T Parameter Kimi

Together AI has integrated Alibaba’s Qwen3.8-27B multimodal model, enabling developers to fine-tune and deploy the model on dedicated infrastructure. The open-weights model outperforms Qwen3.7-Plus and ranks ninth overall in the Code Arena WebDev benchmark with 1,595 points. It is the only model in its size class to place in the top 10, far outperforming the 31 billion parameter Gemma 4-31B, which ranks 80th.

Qwen3.8-27B currently ranks first among open models in the Image-to-WebDev Arena with 1,574 points, matching the performance of Kimi K3, a model with 2.8 trillion parameters. Licensed under Apache 2.0, the model provides a 262,000 native context window and costs $0.40 per million input tokens and $3 per million output tokens.