DeepSeek V3.2 Speciale

DeepSeek

ReasoningMIT

Analysis of DeepSeek's DeepSeek V3.2 Speciale and comparison to other AI models across key metrics including quality, price, performance (tokens per second & time to first token), context window & more.

Context

Release Date: Dec 01, 2025
Window: 128k

PricingPer 1M tokens

Input: $
Output: $
Blended 3:1: $

Capabilities

Input
Output
Reasoning tokens

Benchmarks

Reasoning: ●○○○○
Math: ●●●●●
Coding: ●●○○○
MMLU Pro: 86.3%
GPQA: 87.1%
HLE: 26.1%
SciCode: 44.0%
AIME: 96.7%
LiveCodeBench: 89.6%

Analysis of DeepSeek's DeepSeek V3.2 Speciale and comparison to other AI models across key metrics including quality, price, performance (tokens per second & time to first token), context window & more.