AI Hub
← All models

DeepSeek-V4-Flash vs Hy3

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

  • DeepSeek-V4-Flash has the higher intelligence index (89.4 vs 86.7)
  • Hy3 is cheaper ($0.07 vs $0.10 per 1M input)
  • DeepSeek-V4-Flash is faster
  • DeepSeek-V4-Flash has a larger context window (1M)
DeepSeek-V4-FlashHy3
Intelligence index89.486.7
DeveloperDeepSeekTencent
TypeLLMLLM
AccessOpen weightsOpen weights
Context window1,048,576 tokens262,144 tokens
Input price$0.10 / 1M$0.07 / 1M
Output price$0.20 / 1M$0.26 / 1M
Speed109 tok/s100 tok/s
ReleasedApril 24, 2026April 22, 2026
Parameters284B (13B active)
Input modalitiesTextText
Output modalitiesTextText

Shared benchmarks

DeepSeek-V4-Flash
Hy3
GPQA Diamond
89.4
86.7
Humanity’s Last Exam
32.1
25.5