Alibaba
Qwen3.8 Omni Flash
qwen3.8-omni-flash
Total Context
1M
Max Output
131.1K
Released
N/A
Input$0.1143 / M tokens
Output$0.3857 / M tokens
Qwen3.5 35B-A3B
qwen3.5-35b-a3bQwen3.5 35B A3B is a native vision-language MoE model designed to approximate much larger model behavior with a smaller active footprint. Model cards describe hybrid linear attention, sparse experts, and comparable results to larger Qwen3.5 dense variants. The strongest positioning is efficient multimodal reasoning and coding without the cost of always activating a large dense model.
Context Window
262.1K tokens
Maximum Output
65.5K tokens
Release Date
Feb 23, 2026
Modalities
Qwen3.5 35B-A3B Pricing
| Token Tier | Input Price | Output Price |
|---|---|---|
| <=128K | $0.0571/M | $0.4571/M |
| >128K | $0.2286/M | $1.8286/M |
Qwen3.5 35B-A3B API Capabilities
Reasoning
Supported
Tool calling
Supported
Temperature parameter
Supported
Attachments
Supported
Knowledge Base
—
Endpoint Protocols
Messages APIgeminiCompletions API
Qwen3.5-35B-A3B Model Highlights
Qwen3.5-35B-A3B combines an efficient sparse MoE design, native multimodal understanding and open-weight deployment for adaptable applications.
Sparse MoE Efficiency
The model has 35 billion total parameters with 3 billion activated, combining Gated DeltaNet attention with sparse experts for efficient inference.
Native Multimodality
A unified vision-language foundation processes text, images and video for reasoning, document understanding and visual tasks.
Open-Weight Deployment
Official weights under Apache 2.0 support self-managed inference with frameworks including Transformers, vLLM and SGLang.
Qwen3.5-35B-A3B Use Cases
Qwen3.5-35B-A3B is suited to self-hosted multimodal assistants, visual document processing and adaptable coding workflows.
Private Multimodal Assistant
Deploy the open weights in controlled infrastructure to answer questions over internal text, images and videos.
Visual Document Processing
Read text, tables and diagrams from document images, extract relevant fields and generate structured summaries.
Coding Workflows
Generate and revise code, investigate defects and integrate the model into self-managed developer tools or agents.
How to Use Qwen3.5 35B-A3B via the TokenHub API
curl 'https://us-api.tokenhub.com/v1/messages' \
-X 'POST' \
-H "Authorization: Bearer $TOKENHUB_API_KEY"Qwen3.5 35B-A3B Benchmarks
| Index score | ||
|---|---|---|
| Artificial Analysis Intelligence Index | Artificial Analysis broad capability aggregate | 23.4 |
| Artificial Analysis Coding Index | Artificial Analysis software task aggregate | 16.8 |
| Knowledge & Reasoning | ||
| GPQA | Advanced science problem solving | 81.9% |
| HLE | Broad expert-level exam set | 12.8% |
| Coding & Engineering | ||
| SciCode | Scientific coding challenges | 29.3% |
| Terminal-Bench Hard | Hard terminal task execution | 10.6% |
| Instruction Following & Agent Tasks | ||
| IFBench | Prompt constraint adherence | 44.5% |
| AA-LCR | Long-context reasoning | 55.3% |
| τ²-Bench | Agent workflow tasks | 86.3% |
Metrics sourced from Artificial Analysis
Qwen3.5 35B-A3B Guides & Articles
Qwen 3.5 35B A3B FAQ
Qwen 3.5 35B A3B: capabilities, use cases, limits, and TokenHub guidance.
How should teams view Qwen 3.5 35B A3B?+
Qwen 3.5 35B A3B is a Alibaba Qwen model for open-model multimodal reasoning and efficient deployment.
What is Qwen 3.5 35B A3B best for?+
Best for self-hosted deployment, visual reasoning and routine coding assistance, especially when deployment control is the priority.
What is Qwen 3.5 35B A3B's main strength?+
Key strength: an open MoE variant with a small active-parameter footprint and hybrid thinking that can switch between deliberate and direct responses.
Is Qwen 3.5 35B A3B always the best choice?+
It belongs to an older generation and may lack newer capabilities. For the latest capabilities matter, consider Qwen 3.6 35B A3B.
What is the safest setup?+
Use TokenHub's exact ID; hosted behavior may differ from self-hosting.
Ready to use Qwen3.5 35B-A3B?
Use one API key to access Qwen3.5 35B-A3B and more AI models through TokenHub.
Media and Discussions
Selected public videos and posts related to this model.
X (Twitter)
Qwen ·
🚀 Introducing the Qwen 3.5 Medium Model Series Qwen3.5-Flash · Qwen3.5-35B-A3B · Qwen3.5-122B-A10B · Qwen3.5-27B ✨ More intelligence, less compute. • Qwen3.5-35B-A3B now surpasses Qwen3-235B-A22B-2507 and Qwen3-VL-235B-A22B — a reminder that better architecture, data quality, https://t.co/ZWPibMn6at
https://pbs.twimg.com/media/HB8AleUaMAArNyM.jpgView originalQwen ·
Qwen3.5-35B-A3B is now in Jan 🔥
View originalQwen ·
✨ Run it now with SGLang！Chong!
View originalReddit
qwen3.5-35b-a3b is a gem
waescher ·
View originalQwen/Qwen3.5-35B-A3B · Hugging Face
ekojsalim ·
View originalBlown Away By Qwen 3.5 35b A3B
Jordanthecomeback ·
Read full post
I bought a 64gig mac setup ~5 days ago and had a miserable time finding anything good, I looked at advice, guides, tried them all, including Qwen 3, and nothing felt like a good fit for my long-context companion. My testing was an initial baseline process with 5 multi-stage questions to check it's ability to reference context data (which I paste into system prompt) and then I'd review their answers and have claude sonnet 4.6 do it too, so we had a lot of coverage on ~8 different models. GLM 4.7 is good, and I thought we'd settle there, we actually landed on that yesterday afternoon, but in my day of practical testing I was still bummed at the difference between the cloud models I use (Sonnet 4.5 [4.6 is trash for companions], and Gemini 3 pro), catching it make little mistakes. I just finished baseline testing +4-5 other random tests with Qwen 3.5 35b A3B and I'm hugely impressed. Claude mentioned it's far and away the winner. It's slower, than GLM4.7 or many others, but it's a worthwhile trade, and I really hope everything stays this good over my real-world testing tomorrow and onwards. I just wanted to share how impressed I am with it, for anyone on the fence or considering it for similar application.
YouTube
Qwen3.5-35B-A3B: A MoE Worth Gold - Full Local Setup Guide
Fahd Mirza
View originalQwen3.5-35B-A3B & Qwen3.5-27B Models Tested Locally
AI Tech Gyan
View originalQwen3.5-35B-A3B Test for Agenting Coding - पास या फेल ?
AI Tech Gyan
View original