← All benchmarks

DeepSeek V3.2 Exp (Reasoning)

DeepSeek · Released 2025-09-29 · Independent benchmark scores from Artificial Analysis

Intelligence Index
25.4
Output speed
0.0
tokens/sec
Blended price
$0.32
per 1M tokens
Time to first token
0.00s

Model metadata

Model IDaf134350-8ba3-4629-b56b-00bd6dcf60c4
Slugdeepseek-v3-2-reasoning-0925
Release date2025-09-29
ProviderDeepSeek
Provider slugdeepseek
Output speed0.0 tok/s
Time to first token0.00s
Time to first answer0.00s
Blended price ($/1M)$0.32
Input price ($/1M)$0.28
Output price ($/1M)$0.42

All evaluation scores (17)

aime
aime 2587.7%
artificial analysis coding index
artificial analysis intelligence index25.40
artificial analysis math index87.70
gpqa79.7%
hle13.8%
ifbench54.1%
lcr69.0%
livecodebench78.9%
math 500
mmlu pro85.0%
scicode37.7%
tau banking
tau233.9%
terminalbench hard31.1%
terminalbench v2 1

Source: artificialanalysis.ai/models/deepseek-v3-2-reasoning-0925

Benchmark data from Artificial Analysis.