Artificial Analysis has published scores for Qwen3 Max, the flagship variant of Alibaba's latest open-source model family. The evaluation site, which rates models on a composite of intelligence benchmarks, output speed, and cost-per-million tokens, has added Qwen3 Max to its public leaderboard. Smaller variants - the 27B and 35B parameter versions (where "parameter" count is a rough proxy for model capability and hardware requirements) - are in the queue and have not yet received scores.
Qwen3 is Alibaba's third-generation open-weight model line, released in May 2026. The Max tier sits at the top of the family and targets tasks where raw reasoning quality matters more than speed or cost. For people running models locally or through API providers, Artificial Analysis scores are a practical shortcut: they aggregate multiple benchmarks into a single intelligence rating alongside real-world latency and pricing data, so you can compare models across different hosting providers at a glance.
The 27B and 35B variants are the sizes most relevant to local deployment - they can run on consumer-grade hardware with enough VRAM, unlike the larger Max tier which typically requires server-grade GPUs or cloud API access. Once scored, those results will clarify whether Alibaba's mid-range Qwen3 models deliver competitive reasoning quality at a size that's actually runnable on a workstation.