Alibaba released Qwen3.8-27B under Apache 2.0 this evening: a dense 27-billion-parameter model with native vision, a 262K context window that stretches to 1M, and a published benchmark table showing 61.7 on SWE-bench Pro and 73.0 on Terminal-Bench 2.1. The table itself is the news. The two Qwen3.8 launches before it shipped with no numbers at all.
Read the full story: Qwen3.8-27B ships open weights, scoreboard attached →
Transcript
Alibaba just put Qwen3.8-27B on Hugging Face under Apache 2.0, and this time the benchmarks came with it. It is a dense twenty seven billion parameter model with a vision encoder in the same stack, a two hundred sixty two thousand token context, and it reads image and video natively. The scores: sixty one point seven on SWE-bench Pro, seventy three on Terminal Bench two point one. That puts a model you can run on one accelerator six points behind Alibaba's own two point four trillion parameter flagship. Worth knowing: these are the vendor's own numbers, and this maker's last flagship claimed eleven points higher than an independent harness measured. So treat it as a best case until somebody neutral runs it.