STORY · MODELLER_
Quiet day on the AI front: Qwen3.8-Max launched, DeepSeek Flash impresses
Alibaba presented Qwen3.8-Max with 2.4T parameters and aggressive pricing, with open weights on the way. DeepSeek V4 Flash stood out as the best value model on the Vals Index, while Chinese labs continue to dominate open models. Meanwhile, discussion shows that model quality alone no longer determines agent performance – interplay between model and agent architecture is required.
WHY IT MATTERS
Qwen3.8-Max represents a significant shift from closed to open distribution model at Alibaba. The results show that Chinese labs now lead in open frontier models, and demonstrate that agent development has become as much an infrastructure and design challenge as a reinforcement learning challenge.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.