Qwen3.5-35B-A3B
Alibaba Qwen3.5-35B-A3B is a reasoning vision-language model built on Gated Delta Networks and Mixture-of-Experts, with toggleable thinking and tool use.
Model Overview
Capabilities, design details, and architectural traits
Qwen3.5-35B-A3B - A Reasoning Vision-Language Model With Tool Use
Qwen3.5-35B-A3B is described by the Qwen team as a reasoning vision-language model that supports tool use. It combines multimodal understanding with agentic capability in one model, built on an architecture that favors efficiency over raw size.
| Trait | Detail |
|---|---|
| Defining purpose | A reasoning vision-language model that supports tool use, combining multimodal understanding with agentic capability in one model. |
| Gated Delta plus MoE | Built on a hybrid architecture combining Gated Delta Networks with a sparse Mixture-of-Experts design, aimed at high-throughput inference with low latency. |
| Early fusion training | Trained with early fusion on multimodal tokens, a unified approach to vision-language learning rather than attaching a separate vision encoder to a text only model. |
| Toggleable thinking mode | Supports both a thinking mode for step by step reasoning and a non-thinking mode for direct responses, switchable depending on the task. |
Trained With Massive-Scale Agent Reinforcement Learning
Qwen states it scaled reinforcement learning across environments involving large numbers of agents, using progressively complex task distributions aimed at robust real-world adaptability rather than a single fixed training scenario.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Alibaba | Anthropic | Anthropic |
| Release Date | February 24, 2026 | September 22, 2026 | September 28, 2026 |
| Knowledge Cutoff | - | - | Jun 2026 |
| Context & Limits | |||
| Context Window | 262K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | $0.25 Best Input Pricing | $4 | $2 |
| Output Pricing | $2 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | textimagevideo | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 15.1 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 37.0 | - | - |
| Agentic Index | 11.8 | - | - |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from Alibaba
Other models by Alibaba
Qwen3.8 Max
Qwen3.8 2.4T A95B
Qwen3.8-Flash-Next
Qwen3.8 27B
Top AI Models
Leading alternatives by intelligence score
