VerdictPal
Join

Models / DeepSeek

DeepSeek V3.2 (Reasoning)

DeepSeek V3.2 · Released 2025-12-01

AA Intelligence Index 32.6. Listed on Artificial Analysis with API pricing $0.28/$0.42 per 1M tokens. Benchmark rows sync from the AA snapshot — see the ledger on this page.

Frontier model tracked on Artificial Analysis. Use the benchmark ledger below for sourced scores; we do not infer numbers beyond published rows.

#110 of 166 on AA Index · current snapshot
33 AA Index · 12 benchmark rows · 1 sourcesPosition is benchmark-specific — not a cross-family or cross-source ranking.
Composite61
ContextUnlisted
Input / 1M$0.28
Output / 1M$0.42
Knowledge cutoffunpublished
Statuslightweight

Family profile

Best published score in each covered benchmark family.

78/100 avg
Knowledge
86
Reasoning
84
Math
92
Coding
86
Agentic
47
Long context
71

6 tested benchmark families

Benchmark placements

Where DeepSeek V3.2 (Reasoning) places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.

BenchmarkFamilyRankScoreSourceDate
LiveCodeBenchCoding#9/ 3186Artificial Analysis2026-09-01
AIMEMath#14/ 3792Artificial Analysis2026-09-01
MMLU-ProKnowledge#17/ 3986Artificial Analysis2026-09-01
IFBenchReasoning#80/ 12761Artificial Analysis2026-09-01
AA-LCRLong context#81/ 17971Artificial Analysis2026-09-01
Terminal-BenchAgentic#94/ 18447Artificial Analysis2026-09-01
AA output speedPerformance#102/ 185Artificial Analysis2026-09-01
AA time to first tokenPerformance#102/ 185Artificial Analysis2026-09-01
Humanity's Last ExamReasoning#102/ 18025Artificial Analysis2026-09-01
GPQA DiamondReasoning#110/ 18584Artificial Analysis2026-09-01
Artificial Analysis Intelligence IndexReasoning#128/ 18433Artificial Analysis2026-09-01
SciCodeCoding#137/ 18039Artificial Analysis2026-09-01
Family context

The three highest-scoring models with pages in each capability family. Where DeepSeek V3.2 (Reasoning) shows up, it's highlighted.

KnowledgeDeepSeek V3.2 (Reasoning) · 86
  1. 1
    GPT 5.5
    91
  2. 2
    Claude Opus 4.8
    91
  3. 3
    Claude Sonnet 4.6
    91
ReasoningDeepSeek V3.2 (Reasoning) · 84
  1. 1
    Claude Mythos Preview
    95
  2. 2
    Gemini 3.7 Flash
    95
  3. 3
    Grok 4.6
    95
MathDeepSeek V3.2 (Reasoning) · 92
  1. 1
    OpenAI o3
    99
  2. 2
    GPT-5 (medium)
    99
  3. 3
    GPT-5 (high)
    99
CodingDeepSeek V3.2 (Reasoning) · 86
  1. 1
    GPT 5.5
    96
  2. 2
    Claude Opus 4.8
    96
  3. 3
    Claude Opus 4.6
    96
AgenticDeepSeek V3.2 (Reasoning) · 47
  1. 1
    GPT 5.5
    100
  2. 2
    Claude Opus 4.8
    100
  3. 3
    Claude Fable 5
    100
Long contextDeepSeek V3.2 (Reasoning) · 71
  1. 1
    Muse Spark 1.2
    83
  2. 2
    Kimi K3
    83
  3. 3
    Gemini 3.5 Flash
    81
Newest receipts

Benchmark rows added to the public ledger for DeepSeek V3.2 (Reasoning) in the last 120 days. Older rows live in the full table below.

12 recent
BenchmarkFamilyScoreSourceDays ago
AA-LCRLong context70.67%Artificial Analysis2
AA output speedPerformance0 t/sArtificial Analysis2
AA time to first tokenPerformance0.00sArtificial Analysis2
AIMEMath92%Artificial Analysis2
Artificial Analysis Intelligence IndexReasoning32.8%Artificial Analysis2
GPQA DiamondReasoning84%Artificial Analysis2
Humanity's Last ExamReasoning24.6%Artificial Analysis2
IFBenchReasoning60.68%Artificial Analysis2
LiveCodeBenchCoding86.2%Artificial Analysis2
MMLU-ProKnowledge86.2%Artificial Analysis2
SciCodeCoding38.9%Artificial Analysis2
Terminal-BenchAgentic46.82%Artificial Analysis2
Family coverage

A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.

Knowledge86
Reasoning84
Math92
Coding86
Agentic47
Long context71
Price vs. performance
0255075100$0.10$0.30$1$3$10$30input USD / 1M tokens · log scale →← intelligence

Hover or tab any dot for name, score, and input price.

AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMetaMiniMaxxAIZhipuXiaomiOtherMeituanCohereNVIDIA

Full benchmark ledger

Every catalog benchmark for DeepSeek V3.2 (Reasoning). Scores link to the original source; gaps mean no public row exists yet.

12 sourced rows

12 of 37 catalog benchmarks have a sourced row for DeepSeek V3.2 (Reasoning).

Knowledge

Reasoning

Math

Coding

Agentic

Long context

Performance

Sources