Sobre
DeepSeek-V3.2-Speciale is a high-compute variant of DeepSeek-V3.2 optimized for maximum reasoning and agentic performance. It builds on DeepSeek Sparse Attention (DSA) for efficient long-context processing, then scales post-training reinforcement learning...
Modelos Relacionados
DeepSeek: DeepSeek V4 Pro2026-04-24DeepSeek V4 Pro (Reasoning, High Effort)2026-04-24DeepSeek: DeepSeek V4 Flash2026-04-24DeepSeek V4 Flash (Reasoning, High Effort)2026-04-24DeepSeek V4 Pro (Non-reasoning)2026-04-24DeepSeek V4 Flash (Non-reasoning)2026-04-24DeepSeek: DeepSeek V4 Flash (free)2026-04-24DeepSeek V3.2 (Reasoning)2025-12-01
Preços
Entrada
$0.29
por 1M tokens
Saída
$0.43
por 1M tokens
Combinado
$0.32
por 1M tokens
Mais barato que 61% dos modelos. Preço mediano é $0.54/1M tokens.
Calculadora de Custo
Tokens por dia1M
100K100M
Diário
$0.32
Mensal
$9.69
vs. Modelos Similares
HyperNova 60B 2605Q:-0.1
$0.07-80%
Gemini 2.5 Pro Preview (May' 25)Q:+0.1
$3.44+964%
Qwen3 235B A22B 2507 (Reasoning)Q:+0.1
$0.84+159%
Gemma 4 12B (Reasoning)Q:-0.2
$0.15-54%
Desempenho
Janela de Contexto
164K
tokens
Maior que 48% dos modelos
Saída Máxima
164K
tokens
100% do contexto
Comparação de Janela de Contexto
DeepSeek: DeepSeek V3.1 Terminus
164KIgual
DeepSeek: R1
164KIgual
DeepSeek: R1 0528
164KIgual
Benchmarks
MMLU-Pro
86.3%
GPQA Diamond
87.1%
HLE
26.1%
LiveCodeBench
89.6%
SciCode
44.0%
TerminalBench Hard
34.8%
MATH-500Não avaliado
AIMENão avaliado
AIME 2025
96.7%
IFBench
63.9%
Long Context Recall
59.3%
Tau2
0.0%
Média do MercadoMelhor Score