DeepSeek

DeepSeek V3.2 Exp

by DeepSeektext->text3 endpoints
Neura Intelligence Index
62.8/ 100
Rank
#122
Confidence
medium
Benchmarks
7(18% coverage)

Overview

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...

Capabilities

Text generation
Tool use / Function calling
Structured output
Extended reasoning
Long context
Top-P sampling
Stop sequences
Deterministic seed
Log probabilities

Modalities

Input
Text
Output
Text

Technical Specifications

Context Window
163.8K tokens
Max Output
65.5K tokens
Knowledge Cutoff
2025-07-31
Tokenizer
DeepSeek
Uptime (24h)
9984.1%
Parameters
685000000000B
License
mitOpen Weights
Country
🇨🇳CN
Release Date
September 29, 202511mo

Supported Parameters

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p

Benchmark Performance

general

BenchmarkScoreSource
SimpleQA
97.1
verified

coding

BenchmarkScoreSource
SWE-bench Verified
67.8
verified
TerminalBench
37.7
verified

math

BenchmarkScoreSource
AIME 2025
89.3
verified

reasoning

BenchmarkScoreSource
GPQA Diamond
79.9
verified
Humanity's Last Exam
19.8
verified

agent

BenchmarkScoreSource
BrowseComp
40.1
verified

Performance Over Time

Pricing

Per 1M tokens
Input$0.27
Output$0.41
Added
September 29, 2025
Last synced 9/19/2026