Model specification · DeepSeek

DeepSeek V3.2

deepseek/deepseek-v3.2

DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

SpecificationsReported values
Context window
164K
Maximum output
66K
Architecture
text->text
Tokenizer
DeepSeek
Knowledge cutoff
Moderated
No
Input modalities
text
Output modalities
text
Full pricingUSD where reported
PriceStandard
Input$0.269 / M
Output$0.400 / M
Benchmark indexesReported indexes
Intelligence
32
Coding
44.2
Agentic
18.3
Design Arena Elo
1167
Example workload costsStandard pricing only
WorkloadInput / outputCache shareEstimated cost
Quick check10K / 1K × 10%$0.00309
Long analysis100K / 4K × 150%$0.022
Workday20K / 2K × 1,00030%$5.37
More from DeepSeekAll DeepSeek models
ModelContextInput / MOutput / MIntel.
DeepSeek V4 Flash 07311.0M$0.090$0.18049.9
DeepSeek V4 Pro1.0M$0.435$0.87044.3
DeepSeek V3.1 Terminus164K$0.270$1.0030.4
R1164K$0.700$2.5018.5
DeepSeek V3 0324164K$0.270$1.1215.4