Qwen3.6 35B A3B
Qwen3.6 35B A3B by Alibaba: sparse MoE open-weight model with Gated DeltaNet hybrid, Thinking Preservation, native vision, and agentic coding focus.
Model Overview
Capabilities, design details, and architectural traits
Qwen3.6 35B A3B - Sparse MoE with Thinking Preservation
Qwen3.6 35B A3B is a sparse MoE model (35B total, 3B activated) with a built-in vision encoder. It is the first open-weight release in the Qwen3.6 series. The model card explicitly positions it around two Qwen3.6-specific upgrades: agentic coding (frontend workflows and repository-level reasoning) and Thinking Preservation, a new option to retain reasoning context from historical messages across multi-turn sessions.
| Trait | Detail |
|---|---|
| Thinking Preservation | New in Qwen3.6: optionally retains reasoning traces from prior messages, reducing re-derivation overhead in iterative coding sessions |
| MoE configuration | 256 experts per block; 8 routed + 1 shared expert activated per token |
| Multi-Token Prediction (MTP) | Trained with multi-step MTP, documented in the model card |
| Native multimodal | Includes a vision encoder; supports image, video, and text inputs in a single unified model |
| Context window | 262,144 tokens natively, extensible to 1,010,000 tokens |
| API name | Available as qwen3.6-flash on Alibaba Cloud Model Studio |
| License | Apache 2.0 open-weight |
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 | April 16, 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 | $0.38 Best Input Pricing | $4 | $2 |
| Output Pricing | $2.25 Best Output Pricing | $20 | $10 |
| Modalities | |||
| Inputs | textimagevideo | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 15.2 | 57.6 Best Intelligence Index | 56.0 |
| Coding Index | 28.1 | - | - |
| Agentic Index | 19.2 | - | - |
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
