Qwen3.5 Plus AI Evaluation | LayerLens - Benchmark Results & Performance

Model

Qwen3.5 Plus by Qwen

text+image->text

Free credits included. No credit card required.

The Qwen3.5 native vision-language series Plus models are built on a hybrid architecture that integrates linear attention mechanisms with sparse mixture-of-experts models, achieving higher inference efficiency. In a variety of task evaluations, the 3.5 series consistently demonstrates performance on par with state-of-the-art leading models. Compared to the 3 series, these models show a leap forward in both pure-text and multimodal capabilities.

Evaluations

across 15 benchmarks

Latency

2m 56s

Context length

tokens

Cost

$0.40 · $2.40

input · output per 1M tokens

Key takeaways

Qwen3.5 Plus is a vision-language model (text+image-to-text) designed for state-of-the-art performance in both pure-text and multimodal use cases, particularly excelling in mathematical problem-solving and medical question answering. It aims to solve complex cognitive tasks requiring both textual and visual understanding.
The model is built on a proprietary, hybrid architecture that integrates linear attention mechanisms with sparse Mixture-of-Experts (MoE) models, enhancing inference efficiency. It features a very high context length of 1,000,000 tokens.
Notable capabilities include strong mathematical problem-solving with detailed step-by-step explanations and promising performance in medical question answering, achieving 97% accuracy on the MATH-500 dataset. Limitations include minor algebraic errors, difficulty with specific output formatting, occasional incorrect geometric interpretations, and a complete inability to interpret and execute basic commands or navigate file systems in simulated environments.
Qwen3.5 Plus achieved 97% accuracy on the MATH-500 dataset, demonstrating strong performance in medical question answering and mathematical problem-solving.
The model scored 0% on the SWE-bench Lite (SWE-agent) dataset, indicating a fundamental inability to perform tasks requiring command execution, file system navigation, and Python environment setup within simulated environments, and a lack of logical problem-solving or adaptation to previous failures.

Technical specifications

Context length: 1M tokens
Max output: —
Architecture: MoE
Modalities: text+image->text
Parameters: —
Open weights: No
Licence: Proprietary
Regions: China
Vendor: Qwen
Provider model key: qwen/qwen3.5-plus-2026-02-15