Alibaba Qwen3.6 27B Leads Open-Weight AI Models Under 150B
Alibaba has released two open-weight Qwen3.6 models, with the dense Qwen3.6 27B becoming the highest-scoring open-weight model under 150 billion parameters in reasoning mode on the Artificial Analysis Intelligence Index. The 27B model scored 46, ahead of Qwen3.6 35B A3B at 43, Qwen3.5 27B at 42 and Gemma 4 31B at 39. Alibaba’s open-weight Qwen3.6 lineup remains below 50 billion parameters, while both new models are Apache 2.0 licensed, support 262,000-token context windows and include native multimodal input.
The performance gains come with higher token use and cost. Qwen3.6 27B used about 144 million output tokens to run the Intelligence Index, roughly 3.7 times Gemma 4 31B’s 39 million, translating to an estimated cost of about $659 at Alibaba Cloud pricing versus about $31 for Gemma 4 31B; Qwen3.6 35B A3B cost about $280. On GDPval-AA, Qwen3.6 27B reached 1414 Elo, matching DeepSeek V4 Flash despite that model’s 284 billion total parameters and improving by about 257 Elo over Qwen3.5 27B.
From the sources (3 posts)
@artificialanlysAlibaba's Qwen3.6 27B is the new open weights leader under 150B parameters scoring 46 on the Artificial Analysis Intelligence Index, but uses ~3.7x the output tokens and costs ~21x more than Gemma 4 31B (39) to run the full Intelligence Ind
@artificialanlysQwen3.6 27B used ~144M output tokens to run the Intelligence Index, ~3.7x Gemma 4 31B (39M) for a 7-point higher score. At Alibaba Cloud pricing, this translates to ~$659 for Qwen3.6 27B vs ~$31 for Gemma 4 31B at median third-party pricing
@artificialanlysQwen3.6 27B's GDPval-AA result is notable for its size. At 1414 Elo, it matches DeepSeek V4 Flash (Reasoning, High Effort, 1414) at 284B total parameters and matches Meta’s Muse Spark. This represents a ~257 Elo gain over Qwen3.5 27B (1157)