Qwen3.5-397B-A17B
Active Parameters
397B
Context Length
262K
Modality
Multimodal
Auxiliary Parameters
450M
Architecture
Mixture of Experts
KV Cache Rate
120 KB/tok
License
Apache 2.0
Release Date
24 Feb 2026
Knowledge Cutoff
-
API Pricing (per 1M)
Input: $0.50 · Output: $3.60
Hardware Sizing
Precision
Context
Workload Memory: 774.57 GB
FP16 · 8K context
Base Weights: 771.14 GB
KV Cache: 0.94 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
8x NVIDIA H200 141GB
774.6 / 1015.2 GB usable · TP 8
Determine requirements for datacenter hardware:
Upgrade to Pro
Architecture Diagram
Evaluation Benchmarks
Rank
#52
| Benchmark | Score | Rank |
|---|---|---|
Professional Knowledge | 0.878 | 4 |
StackUnseen | 0.763 | 16 |
Graduate-Level QA | 0.884 | 29 |
General Text | 1442 | 73 |
Web Development | 1400 | 81 |
Coding Index | 0.48 | 91 |
Agentic Index | auto 0.08 | 94 |
Intelligence Index | auto 0.18 Standard 0.21 | 175 142 |
Rankings
Overall Rank
#52
Coding Rank
#85
About Qwen3.5-397B-A17B
Qwen3.5-397B-A17B is Alibaba Cloud's largest and most capable multimodal foundation model, released February 2026. With 397B total parameters and 17B activated through a Mixture-of-Experts architecture (512 experts), it achieves state-of-the-art scores on MMLU-Pro (87.8%), GPQA Diamond (88.4%), SWE-bench Verified (80.0%), and Terminal-Bench 2.0 (54.0%). It features unified vision-language capabilities, extended context up to 1M tokens, and excels in coding agents, general agents, multimodal reasoning, and multilingual understanding across 201 languages.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
32
Key-Value Heads
2
Attention Head Dimension
256
Position Embedding
ROPE
RoPE Theta
10,000,000
Linear Attention
Yes
Linear Attention Ratio
75.0%
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Auxiliary Parameters
450M
Hidden Dimension Size
4,096
Number of Layers
60
FFN Intermediate Size (Dense)
1,024
Multi-Token Prediction Heads
1
Tokenizer
Vocabulary Size
248,320
Mixture of Experts
Total Expert Parameters
17.0B
Number of Experts
512
Active Experts
11
FFN Intermediate Size (per Expert)
1,024
About Qwen 3.5
Qwen 3.5 is Alibaba Cloud's latest-generation foundation model family, released February 2026. It represents a significant leap forward, integrating breakthroughs in multimodal learning (unified vision-language foundation), efficient hybrid architecture (Gated Delta Networks with sparse Mixture-of-Experts), scalable reinforcement learning across million-agent environments, and global linguistic coverage spanning 201 languages. Available under Apache 2.0 license with open weights.