Qwen3.5 Plus

An independent technical guide from February 16, 2026 discusses Qwen3.5-Plus alongside the open-weight Qwen3.5-397B-A17B, presenting long-context, tool integration, and API convenience versus hardware and control advantages of self-deployment. It includes a runnable OpenAI-compatible SDK example and covers deployment v The article lists benchmark scores for the Qwen3.5 series: AIME26 91.3, GPQA Diamond 88.4, LiveCodeBench v6 83.6, SWE-bench Verified 76.4, and Terminal-Bench 2 52.5 (plus additional BFCL results), comparing the series against GPT-5.2, Claude Opus 4.5, and Gemini 3 Pro. It explicitly notes that item prompts, seeds, repe

The wiki entry explicitly distinguishes Qwen3.5-Plus as the hosted proprietary model in Alibaba's Qwen 3.5 family, paired with the open-weight Qwen3.5-397B-A17B. It notes a reported 1M-token context window for the hosted Qwen3.5 Plus offering, highlighted in February 18, 2026 newsletter coverage, alongside multimodal s For AI product managers, the entry frames Qwen3.5-Plus as relevant to evaluating hosted long-context models for copilots, agents, and developer tools, given its up to 1M-token hosted context and agent-friendly capabilities. Key developments tracked include Simon Willison's February 17, 2026 review distinguishing the ho

On February 16, 2026 (Chinese New Year's Eve), Alibaba released Qwen3.5-Plus as the hosted proprietary counterpart in its Qwen 3.5 family, positioned as comparable to Gemini 3 Pro and topping open-source rankings. According to the supplied article, the model uses a 397-billion-parameter MoE architecture with only 17 bi The article reports benchmark scores including 87.8 on MMLU-Pro (surpassing GPT-5.2), 88.4 on GPQA (outperforming Claude 4.5), and a record on IFBench, claiming Qwen3.5-Plus comprehensively surpasses Gemini 3 Pro. Unlike previous text-only generations, it achieves native multimodality through early text-vision fusion p

Qwen3.5 is described as a flagship large model series launched by Alibaba on February 16, 2026, comprising two versions: Qwen3.5-Plus and Qwen3.5-397B-A17B. The series supports text and multimodal tasks using a Hybrid Attention Mechanism and Sparse Mixture of Experts (MoE) architecture, optimized for logical reasoning, The Qwen3.5-Plus API price is 0.8 RMB per million tokens (reportedly 1/18th of Gemini 3 Pro's price), and it supports native multimodal training with an MMLU-Pro knowledge reasoning benchmark score of 87.8. Visual capabilities cover multimodal reasoning, text recognition, and video understanding. Through its hybrid arc

The Baidu encyclopedia entry confirms Qwen3.5-Plus as an open-source large language model released by Alibaba on February 16, 2026, belonging to the new generation of the Qianwen series. It is based on the Qwen3-Next architecture combining higher-sparsity MoE with a hybrid attention mechanism of Gated DeltaNet and Gate Additional specifications include expanded language support from 119 to 201 languages and vocabulary growth from 150,000 to 250,000 tokens, with benchmark performance comparable to closed-source models like GPT-5.2 and Gemini-3-Pro. The entry documents that on February 17, 2026 (the day after release), Huawei Computing

Alibaba officially released the Qwen3.5 series on February 16, 2026, open-sourcing Qwen3.5-397B-A17B — which Alibaba itself states is "also named Qwen3.5-Plus." The model is described as a natively multimodal foundation model trained on trillions of vision-language tokens across multilingual text, images, videos, STEM, Qwen3.5-Plus combines a linear attention mechanism with a sparse mixture-of-experts (MoE) design to reduce compute requirements, activating only 17 billion of its 397 billion total parameters. Across benchmarks, it demonstrates strong performance in language understanding and reasoning, code generation, agentic workflo