DeepSeek V3.2
by DeepSeekRank #70Score 83.4
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...
Performance Overview
Score
83.4
Rank
#70
24h Change
-1
7d Change
-6
State
stableConfidence
highSignal Scores
| Signal | Normalized | Weight | Contribution | Freshness |
|---|---|---|---|---|
Capabilities capability | 66.7 | 20% | 13.3 | 2026-09-22T14:23:19.816Z |
Benchmarks benchmark | 86.7 | 30% | 26.0 | 2026-09-22T14:23:19.816Z |
Pricing pricing_tier | 99.6 | 15% | 14.9 | 2026-09-22T14:23:19.816Z |
Context Window context_window | 82.8 | 10% | 8.3 | 2026-09-22T14:23:19.816Z |
Recency recency | 79.1 | 15% | 11.9 | 2026-09-22T14:23:19.816Z |
Output Capacity output_capacity | 77.0 | 10% | 7.7 | 2026-09-22T14:23:19.816Z |
Top Drivers
positive
Benchmarks
positive
Pricing
$0.40/M output tokens
$0.40
positive
Capabilities
Supports reasoning, tools, JSON mode, streaming
4/7
positive
Recency
Released over 1 year ago
10mo ago
Capabilities
| Capability | Supported |
|---|---|
| Vision | No |
| Reasoning | Yes |
| JSON Mode | Yes |
| Streaming | Yes |
| Function Calling | Yes |
| Web Search | No |
Pricing
Input / 1M tokens
$0.27
Output / 1M tokens
$0.40
Context Window
164K
Max Output
66K