Model specification · Qwen

Qwen3.5 397B A17B

qwen/qwen3.5-397b-a17b

The Qwen3.5 series 397B-A17B native vision-language model is built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. It delivers...

SpecificationsReported values
Context window
262K
Maximum output
236K
Architecture
text+image+video->text
Tokenizer
Qwen3
Knowledge cutoff
—
Moderated
No
Input modalities
text image video
Output modalities
text
Full pricingUSD where reported
PriceStandard
Input$0.550 / M
Output$3.50 / M
Benchmark indexesReported indexes
Intelligence
19.1
Coding
48.2
Agentic
10.6
Design Arena Elo
1183
Example workload costsStandard pricing only
WorkloadInput / outputCache shareEstimated cost
Quick check10K / 1K × 10%$0.00900
Long analysis100K / 4K × 150%$0.053
Workday20K / 2K × 1,00030%$16.05
More from QwenAll Qwen models
ModelContextInput / MOutput / MIntel.
Qwen3.8 Max (0902)1M$2.00$6.0045.4
Qwen3.8 2.4T A95B1.0M$2.00$6.0040
Qwen3.8 27B1M$0.214$2.5533.9
Qwen3.7 Max1M$1.48$4.4229.9
Qwen3.7 Plus1M$0.320$1.2825.8
Qwen3.6 27B262K$0.300$2.0021.9
Qwen3.6 35B A3B262K$0.100$0.90018.8
Qwen3.5-122B-A10B262K$0.260$2.0816.2
Qwen3 235B A22B Thinking 2507131K$0.230$2.3012.7
Qwen3 Coder Next262K$0.120$0.80010.1
Qwen3 30B A3B Thinking 250782K$0.200$2.409.8
Qwen3 32B131K$0.080$0.2807.2