Agent Mag
SUBSCRIBE
Qwen3.5-35B-A3B
All models
AlibabaAlibabaQwenReleased 2026-02-25

Qwen3.5-35B-A3B

262K context$0.163/M input$1.30/M output35B

The Qwen3.5-35B-A3B is a vision-language model with a hybrid architecture combining linear attention mechanisms and a sparse mixture-of-experts model. It is designed for higher inference efficiency and achieves performance comparable to the Qwen3.5-27B. The model supports reasoning-enabled tasks and structured outputs, making it suitable for complex applications.

What is Qwen3.5-35B-A3B?

Qwen3.5-35B-A3B is an AI model from Alibaba that Agent Mag tracks for pricing, context window, modalities, benchmarks, and API compatibility. Builders can use this page to compare Qwen3.5-35B-A3B against other models for agent workflows and production deployments.

Model ID

The Qwen3.5-35B-A3B is a vision-language model with a hybrid architecture combining linear attention mechanisms and a sparse mixture-of-experts model. It is designed for higher inference efficiency and achieves performance comparable to the Qwen3.5-27B. The model supports reasoning-enabled tasks and structured outputs, making it suitable for complex applications.

Architecture & Specifications
Architecture
Hybrid architecture with linear attention and sparse mixture-of-experts
Parameters
35B
Tokenizer
Qwen3
License
Standard
Released
2026-02-25
Modalities
Input
textimagevideo
Output
text
Supported Parameters
frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Strengths
  • Hybrid architecture for higher inference efficiency
  • Supports reasoning-enabled tasks
  • Structured output capabilities
  • Vision-language integration
  • Comparable performance to Qwen3.5-27B
Limitations
  • Low performance on research-level physics reasoning (CritPt: 0.9%)
  • Moderate hallucination rate in knowledge tasks (16.0%)
  • Limited accuracy in economically valuable tasks (GDPval-AA: 20.6%)
  • Lower coding capability compared to specialized models
  • Performance varies across benchmarks
Recommended Use Cases
Scientific reasoning tasks
Vision-language applications
Instruction-following scenarios
Agentic coding workflows
Long context reasoning evaluations

More from Alibaba

Qwen3.6 Plus
1M ctx$0.325/M

Qwen 3.6 Plus builds on a hybrid architecture that combines efficient linear attention with sparse mixture-of-experts routing, enabling strong scalability and high-performance inference. Compared to the 3.5 series, it delivers...

Qwen3.5-9B
262K ctx$0.100/M

Qwen3.5-9B is a multimodal foundation model from the Qwen3.5 family, designed to deliver strong reasoning, coding, and visual understanding in an efficient 9B-parameter architecture. It uses a unified vision-language design...

Qwen3.5-122B-A10B
262K ctx$0.260/M

The Qwen3.5 122B-A10B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. In terms of...

Qwen3.5-27B
262K ctx$0.195/M

The Qwen3.5 27B native vision-language Dense model incorporates a linear attention mechanism, delivering fast response times while balancing inference speed and performance. Its overall capabilities are comparable to those of...

Related content

Browse the AI models directory

Compare pricing, local installs, context windows, and modality filters across the full model catalog.

Explore AI agent tools

Find frameworks, SDKs, and infrastructure tools that pair with this model in production workflows.

Read Agent Mag model and agent articles

See Agent Mag coverage of model benchmarks, agent frameworks, and deployment patterns.

Data enriched Apr 24, 2026. Pricing from OpenRouter API.