ModelsCompareBest forBenchmarksStatusPricingAPI
Compare / head-to-head

DeepSeek V3.2vsGemma 4 31B

Gemma 4 31B leads 3 of 5 shared benchmarks. Gemma 4 31B has the larger context window (256K tokens).

Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.

Shared benchmarks
2 – 3
Gemma 4 31B leads
Cheaper per token
list price, input + output
Larger context
Gemma 4 31B
256K tokens
Provider uptime (30d)
100% · 100%
DeepSeek · Google
DeepSeek V3.2
DeepSeek · released 2025-12-01
text
Context
131K
Max out
66K
Input /1M
Output /1M
Cached /1M
Scores
8 · 7 core
Gemma 4 31B
Google · released 2026-04-02
textvision
Context
256K
Max out
Input /1M
Output /1M
Cached /1M
Scores
6 · 5 core
Quality

Benchmark matrix

BenchmarkDeepSeek V3.2Gemma 4 31BΔEdge
Reported by both · 5
GPQA Diamond82.4% 84.3% -1.9 ptGemma 4 31B
MMLU-Pro85% 85.2% -0.2 ptGemma 4 31B
AIME 202694.2% matharena89.2% +5 ptDeepSeek V3.2
Humanity's Last Exam (no tools)25.1% 19.5% +5.6 ptDeepSeek V3.2
LMArena Elo1424.8 1441.7 -16.9Gemma 4 31B
Only DeepSeek V3.2 reports · 2
SWE-bench Verified73.1% not reported
AIME 202593.1% not reported
Only Gemma 4 31B reports · 1
LiveCodeBench v6not reported80%
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is DeepSeek V3.2 minus Gemma 4 31B in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing

Side by side

SpecDeepSeek V3.2Gemma 4 31BEdge
Context window131K tokens256K tokensGemma 4 31B
Max output66K tokens
Input price / 1M
Output price / 1M
Cached input / 1M
Input + output / 1M
Lower is cheaper. List prices; batch, tool and regional fees excluded.
Modalitiestexttext · visionGemma 4 31B
Released2025-12-012026-04-02
Cited benchmark scores86
Reliability

Provider status

All providers
Built by Respan
Which one wins on your data?

Public benchmarks are a starting point. Run DeepSeek V3.2 and Gemma 4 31B on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.