Skip to content
SOLVENCY

Head to head

DeepSeek V4 Flash vs Grok 4.5

DeepSeek V4 Flash: currentGrok 4.5: currentheavy tier

Verdict · measured basis

DeepSeek V4 Flash costs 32x less per solved task.

On output token price alone, DeepSeek V4 Flash looks 4.5x cheaper (4.5x on input). The ranking holds, but the margin changes.

Source: Artificial Analysis (artificialanalysis.ai) · verified 2026-08-21 · prices from each provider's pricing page

Input $/M tokens

lower is better

  • DeepSeek V4 Flash$0.44
  • Grok 4.5$2

Output $/M tokens

lower is better

  • DeepSeek V4 Flash$1.32
  • Grok 4.5$6

Cost per solved task

measured basis · lower is better

  • DeepSeek V4 Flash$0.12
  • Grok 4.5$3.81
DeepSeek V4 FlashGrok 4.5
Input $/Mtok$0.44$2
Output $/Mtok$1.32$6
Cached input $/Mtok$0.014$0.3
Pass rate50%64%
Pass rate sourceAAAA
Cost basismeasuredmeasured
Cost per solved task$0.12$3.81

verified 2026-08-21 · heavy task tier; measured rows ignore the tier· Methodology· DeepSeek V4 Flash· Grok 4.5