Alibaba's retrained model now tops a front-end coding leaderboard.
Alibaba retrained its Qwen3.8-Max model on 2 September without changing its size or design. It scored 1,691 on the CodeArena front-end chart, three points above Claude Opus 5. It now runs in Alibaba's API and apps at about a quarter of the second model's price.
Why it mattersCheap coding models decide what your tools cost, and this one leads while charging less. Post-training gains mean labs can climb charts without buying more chips.
Read in the app — free, in 9 languages
Related stories
A startup whose agents reset passwords sold for $500 million.2026-09-03A startup that coordinates other firms' AI agents hit $5 billion.
2026-09-03A startup that guards AI models grew its revenue tenfold.
2026-09-03Meta's new coding model tells you when it is stuck.
2026-09-03Google's small new model finds real security holes 70% of the time.
2026-09-03