Head to head

DeepSeek V4 Pro vs Gemma 4 31B Instruct (Pearl AI)

On a blended workload of one million input tokens and three million output tokens, Gemma 4 31B Instruct (Pearl AI) is 1.1× cheaper than DeepSeek V4 Pro.

DeepSeek V4 ProGemma 4 31B Instruct (Pearl AI)
Provider DeepSeek Together AI
Input / MTok $0.435 $0.28
Output / MTok $0.87 $0.86
Cache read / MTok $0.003625
Context window 1,000,000 32,000
Blended 1M in + 3M out $3.04 $2.86
Status ga ga

DeepSeek V4 Pro detail   Gemma 4 31B Instruct (Pearl AI) detail