DeepSeek-V3.2 (Thinking) vs Gemini 3 Flash: Specs & Benchmark Comparison

CharacteristicDeepSeek-V3.2 (Thinking)Gemini 3 Flash
CompanyDeepSeekGoogle
Release DateNovember 30, 2025December 16, 2025
Parameters685B
MultimodalNoYes
Context (input)131K1.0M
Context (output)66K66K
Input Price / 1M$0.28$0.50
Output Price / 1M$0.42$3.00
Average Score0.90.9
Benchmarks
GPQA0.80.9
AIME 20250.91.0

Visual Benchmark Comparison

DeepSeek-V3.2 (Thinking)
Gemini 3 Flash
GPQA0.8 vs 0.9
0.8
0.9
AIME 20250.9 vs 1.0
0.9
1.0

Verdict

Gemini 3 Flash leads in 2 out of 4 comparison categories.

Overall Performance

Both models show comparable average scores: DeepSeek-V3.2 (Thinking) — 0.9, Gemini 3 Flash — 0.9.

API Cost

DeepSeek-V3.2 (Thinking) is 5.0x cheaper: input $0.28/1M vs $0.50/1M tokens.

Context Window

Gemini 3 Flash supports a larger context: 1M vs 131K tokens.

Recency

Gemini 3 Flash is newer: released 12/16/2025 vs 11/30/2025.

More About These Models

Related Comparisons

The DeepSeek-V3.2 (Thinking) and Gemini 3 Flash comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the DeepSeek-V3.2 (Thinking) or Gemini 3 Flash page. See also the complete list of AI model comparisons.