token.app › Compare › DeepSeek V3.2 vs Gemma 4 31B

DeepSeek V3.2 vs Gemma 4 31B

Gemma 4 31B leads on 2 of 2 shared benchmarks. Gemma 4 31B is 1.8× cheaper per token on a blended 3:1 basis. Gemma 4 31B accepts the larger context window (262K).

Head to head
Blended price uses a 3:1 input:output mix
DeepSeek V3.2Gemma 4 31B
ProviderDeepSeekGoogle
Input $/1M$0.260$0.100
Output $/1M$0.380$0.340
Blended $/1M$0.290$0.160
Context window164K262K
Max output164K262K
ReleasedDec 2025Apr 2026
LicenceOpen weightsOpen weights
GPQA Diamond 71.2% 75.8%
OTIS Mock AIME 48.9% 73.3%
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose DeepSeek V3.2 if…
  • No clear advantage on the data we hold.
Choose Gemma 4 31B if…
  • Costs less per token — $0.160 vs $0.290 blended
  • Higher GPQA Diamond (75.8% vs 71.2%)
  • Higher OTIS Mock AIME (73.3% vs 48.9%)
  • Larger context window (262K)
  • Newer model (released Apr 2026)
FAQ
Which is better, DeepSeek V3.2 or Gemma 4 31B?
Across the 2 benchmarks both models have independently-run scores for, DeepSeek V3.2 leads on 0 and Gemma 4 31B leads on 2. Benchmarks are one input — price and context window are on this page too.
Is DeepSeek V3.2 cheaper than Gemma 4 31B?
Gemma 4 31B is cheaper: $0.160 versus $0.290 per million tokens blended at 3:1 input:output. Input alone: $0.260 vs $0.100. Output alone: $0.380 vs $0.340.
What context windows do DeepSeek V3.2 and Gemma 4 31B have?
DeepSeek V3.2: 163,840 tokens. Gemma 4 31B: 262,144 tokens.
Who makes DeepSeek V3.2 and Gemma 4 31B?
DeepSeek V3.2 is made by DeepSeek. Gemma 4 31B is made by Google.
Related comparisons