Alibaba's open Qwen 3.5 takes aim at GPT-5 mini and Claude Sonnet 4.5 at a fraction of the cost

2026-02-26 · Source: The Decoder · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, quick

Summary

Alibaba has released an expanded Qwen 3.5 model series, including Qwen3.5-Flash, Qwen3.5-35B-A3B, Qwen3.5-122B-A10B, and Qwen3.5-27B. These models accept text, images, and video as input, generating text output, and are designed for stronger performance with reduced computational requirements. The smaller Qwen3.5-35B-A3B notably surpasses its larger predecessor, Qwen3-235B-A22B, indicating architectural and data quality improvements. Benchmarks show Qwen 3.5 models matching or exceeding Western counterparts like OpenAI's GPT-5 mini and Anthropic's Claude Sonnet 4.5 in areas such as agent-based tool use (BFCL V4, 72.2), web search (BrowseComp, 63.8), visual reasoning (MMMU-Pro, 76.9), and document recognition (OmniDocBench, 89.8). The models are available on Hugging Face and ModelScope under the Apache License 2.0, with Qwen3.5-Flash offering a 1-million-token context length and an API cost of $0.10 per million input tokens and $0.40 per million output tokens.

Key takeaway

For NLP Engineers and CTOs evaluating cost-effective, high-performance multimodal LLMs, Alibaba's Qwen 3.5 series presents a compelling open-source alternative. Its competitive benchmark scores against GPT-5 mini and Claude Sonnet 4.5, coupled with an Apache 2.0 license and low API costs for Qwen3.5-Flash, suggest a viable option for deploying advanced AI capabilities without the premium associated with proprietary models. Consider integrating Qwen 3.5 for agent-based applications requiring multimodal input.

Key insights

Alibaba's Qwen 3.5 models achieve competitive performance against top Western LLMs at lower computational cost.

Principles

• Architecture and data quality outweigh raw model size.
• Open-source models can rival proprietary top-tier LLMs.

In practice

• Deploy Qwen 3.5 models via Hugging Face or ModelScope.
• Utilize Qwen3.5-Flash for production with 1M token context.
• Integrate Qwen 3.5 for multimodal agent-based applications.

Topics

• Alibaba Qwen 3.5
• Multimodal AI Models
• Large Language Models
• Open-source AI
• AI Benchmarking

Best for: NLP Engineer, Computer Vision Engineer, CTO, AI Engineer, Machine Learning Engineer, AI Product Manager

Related on AIssential

• Alibaba just open-sourced a model that rivals GPT-5.2 — Alibaba's Qwen 3.5, with 17 billion active parameters, rivals top closed-source models while being open-weight.
• Alibaba's new open source Qwen3.5-Medium models offer Sonnet 4.5 performance on local computers — Alibaba's Qwen3.5 models offer high-performance, open-source LLMs with efficient local deployment and competitive API pricing.
• Qwen 3.5 Small Models: Alibaba Delivers Compact Multimodal Powerhouses — Alibaba's Qwen 3.5 Small models offer powerful multimodal AI for edge devices.
• Alibaba's Qwen 3.5 397B-A17 beats its larger trillion-parameter model — at a fraction of the cost — Sparse Mixture-of-Experts models can surpass larger predecessors in performance and efficiency.
• Alibaba launches Qwen 3.5 AI model amid intensifying China chatbot race — Alibaba's Qwen 3.5 model offers significant cost reduction and workload capacity for agentic AI tasks.
• 🗞️ Alibaba released their open-source Qwen3.5-9B model, and it holds its own against OpenAI's 120B param gpt-oss system. — Smaller, efficient AI models are achieving performance comparable to much larger systems through architectural innovations and multimodal training.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Decoder.