NEWHow AI models draw the world map · Gecko Tests
Beta
Home/Models/DeepSeek V3.2 Exp
DeepSeek logo

DeepSeek V3.2 Exp

by DeepSeek · Released Sep 2025

Open Source
51.6
avg score
Rank #133
Compare
Better than 57% of all models
Context
164K tokens (~82 books)
Input $/1M
$0.27
Output $/1M
$0.41
Type
text
License
Open Source
Benchmarks
18 tested
Data as of
About

DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

Tested on 18 benchmarks · BenchGecko score 51.6. Top scores: Chatbot Arena Elo — Overall (1422.1%), Chatbot Arena Elo — Coding (1271.8%), Fiction.LiveBench (83.3%).

Looking for similar performance at lower cost?
MiniMax M2.7 scores 51.2 (99% as good) at $0.21/1M input · 22% cheaper
Capabilities
coding
55.9
#65 globally
reasoning
44.9
#90 globally
math
64.4
#64 globally
knowledge
66.6
#30 globally
agentic
42.9
#27 globally
language
42.5
#131 globally
general
34.1
#110 globally
Benchmark Scores
Compare All
Tested on 18 benchmarks · Ranked across 8 categories
Score Distribution (all 312 models)
0255075100
▲ You are here
codingCompare coding →
Aider polyglot

Multi-language code editing from Aider. Tests editing ability across Python, JavaScript, TypeScript, Java, C++, Go, Rust, and more.

74.2·
LiveBench — Coding

Regularly refreshed coding problems that avoid data contamination. New problems added monthly to prevent memorization.

73.2·
WeirdML

Unusual and adversarial machine learning challenges. Tests robustness of reasoning about edge cases in ML systems.

39.5·
reasoningCompare reasoning →
LiveBench — Reasoning

Regularly refreshed reasoning problems testing logical deduction, spatial reasoning, and analytical thinking.

45.5·
LiveBench — Data Analysis

Fresh data analysis tasks testing ability to interpret tables, charts, and statistical data.

44.3·
mathCompare math →
LiveBench — Mathematics

Regularly updated math problems that test numerical reasoning, algebra, calculus, and combinatorics.

64.4·
Quick compare:z-ai logovs GLM 4.7DeepSeek logovs DeepSeek V3.1minimax logovs MiniMax M2.5
Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Model Family · DeepSeek DeepSeek V3.2
Dec 2025
51.3
$0.28/M in164Kctx35 benchmarks
Sep 2025
51.6+0.3
$0.27/M in(-0.01)164Kctx18 benchmarks
Dec 2025
95.2+43.6
$0.40/M in(+0.13)164Kctx9 benchmarks
See the full DeepSeek V3.2 family →
Recently Happened
DeepSeek V3.2 Exp pricing increased 103%
Sep 30, 2026
Similar Models
z-ai logo
GLM 4.7
z-ai
51.7$0.60/1M
DeepSeek logo
DeepSeek V3.1
DeepSeek
51.8$0.25/1M
minimax logo
MiniMax M2.5
minimax
51.4$0.27/1M
Links
Info
DeepSeekPricing explorerDevelopers · API
Research
Technical Report
Documentation
API DocsPlayground
Community
@DeepSeek
Source Code
GitHubHugging Face
BenchGecko API
deepseek-v3-2-exp
Specifications
  • Typetext
  • Context164K tokens (~82 books)
  • ReleasedSep 2025
  • LicenseOpen Source
  • StatusActive
  • Cost / Message~$0.001
Available On
DeepSeek logoDeepSeek$0.27
Categories
codingreasoningmathknowledgeagenticlanguagegeneral
Learn More
context-windowtransformeropen weightstokens
Share & Export
Tweet
Related Models
z-ai logoGLM 4.7
51.7
DeepSeek logoDeepSeek V3.1
51.8
minimax logoMiniMax M2.5
51.4
OpenAI logoo4 Mini
51.4
DeepSeek V3.2 Exp is an open-source text AI model by DeepSeek, released in September 2025. It has an average benchmark score of 51.6. Context window: 164K tokens.

Related Models

GLM 4.7 · z-aiDeepSeek V3.1 · DeepSeekMiniMax M2.5 · minimaxo4 Mini · OpenAIDeepSeek V3.2 · DeepSeek

Benchmarks

Chatbot Arena Elo — OverallChatbot Arena Elo — CodingFiction.LiveBenchDtbenchAider polyglot

Related Pages

DeepSeek · ProviderDeepSeek · EconomyAll ModelsCompare ModelsPricingDevelopers · APIContext Window · Glossary

Key facts · as of 2026-10-05

  • DeepSeek V3.2 Exp by DeepSeek. BenchGecko score 51.6, rank 133 of 312 scored models (normalized average of public benchmark scores).
  • List price $0.27 input · $0.41 output per 1M tokens (as of 2026-10-05).
  • Sold by 1 provider (as of 2026-10-05): SiliconFlow (fp8) $0.27 in / $0.41 out.

How to cite · data as of 2026-10-05

DeepSeek V3.2 Exp · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-05. https://benchgecko.ai/model/deepseek-v3-2-exp

Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP