AI Hub
← All models

DeepSeek V3.2 Speciale vs Llama 3.3 70B Instruct

DeepSeek vs Meta — benchmarks, pricing, and capabilities side by side.

  • DeepSeek V3.2 Speciale has the higher intelligence index (89.9 vs 50.6)
  • Llama 3.3 70B Instruct is cheaper ($0.10 vs $0.29 per 1M input)
  • DeepSeek V3.2 Speciale has a larger context window (164K)
DeepSeek V3.2 SpecialeLlama 3.3 70B Instruct
Intelligence index89.950.6
DeveloperDeepSeekMeta
TypeLLMLLM
AccessOpen weightsOpen weights
Context window163,840 tokens131,072 tokens
Input price$0.29 / 1M$0.10 / 1M
Output price$0.43 / 1M$0.32 / 1M
Speed2220 tok/s
ReleasedDecember 1, 2025December 6, 2024
Parameters
Input modalitiesTextText
Output modalitiesTextText

Shared benchmarks

DeepSeek V3.2 Speciale
Llama 3.3 70B Instruct
AIME 2025
96.7
7.7
GPQA Diamond
87.1
50.5
Humanity’s Last Exam
26.1
4
LiveCodeBench
89.6
28.8
MMLU-Pro
86.3
68.9