Back to the index
DeepSeek · DeepSeek
No. 26

DeepSeek-V3.2

Ultra-cheap open MoE with standout math and efficiency.

OpenOpen weightsText
Field notes
Context window
131Ktokens
Max output
66Ktokens
Input price
$0.23/ 1M tokens
Output price
$0.34/ 1M tokens
Cached input
$0.03/ 1M tokens
Modalities
Text
Knowledge cutoff
Jul 2025
Released
Sep 2025
Parameters
671B
License
MIT
Approx. speed
Overview

What this model is

DeepSeek-V3.2 is DeepSeek's V3-generation open-weight chat model (671B MoE, 37B active, MIT), which introduced DeepSeek Sparse Attention for efficient long-context inference. It delivers strong math and general reasoning at industry-low prices, with famously cheap cache hits. Now a generation behind the V4 line, it remains a high-value text-only workhorse rather than a frontier agentic coder.

Strengths

  • Among the lowest prices in the industry
  • Excellent math and reasoning value
  • Sparse-attention efficiency at long context
  • Very cheap cached input; MIT open weights

Trade-offs

  • Superseded by the V4 generation on price and capability
  • Agentic-coding reliability below the frontier
  • Text-only, no vision
  • English/Chinese-centric multilingual

Best for

  • High-volume, cost-sensitive workloads
  • Math and reasoning tasks
  • Chat and summarization at scale
  • Self-hosted deployments

Not ideal for

  • Long-horizon autonomous agentic coding
  • Multimodal tasks
Capabilities

Capability profile

Normalized 0–100 scores, comparable across the whole catalog.

Reasoning80Coding63Math88Writing78Knowledge82Speed70Agentic60Vision5Multilingual80Long Context76
Reasoning
80
Coding
63
Math
88
Writing
78
Knowledge
82
Speed
70
Agentic
60
Vision
5
Multilingual
80
Long Context
76
Benchmarks

How it scores

Public benchmark results, with independent third-party results where available. Bars normalize percentages to 100 and Elo ratings to a 1500 ceiling.

Independent index
Artificial Analysis Intelligence Index
Composite of ~9–10 independent evals · Artificial Analysis
25/ 100
BenchmarkResult
Artificial Analysis Intelligence Indexindependent
GeneralArtificial Analysis
25%
SWE-bench Verifiedindependent
CodingSWE-bench
73%
SWE-bench Proindependent
CodingSWE-bench Pro
15.6%
Aider Polyglotindependent
CodingAider
74.2%
τ-benchindependent
Agenticτ-bench
71.3%
MMLU-Pro
KnowledgeVendor-reported
85%
GPQA Diamondindependent
ReasoningIndependent aggregators
80%
AIME 2025
MathVendor-reported
89.3%
LiveCodeBenchindependent
CodingLiveCodeBench
49.2%
LMArena Eloindependent
GeneralLMArena
1455Elo
Alternatives

Comparable models

Models in a roughly similar class — worth weighing against this one.