Codesota · Models · DeepSeek-V3.2DeepSeek14 results · 8 benchmarks
Model card

DeepSeek-V3.2.

DeepSeekopen-source
§ 02 · Benchmarks

Every benchmark DeepSeek-V3.2 has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01AIME 2025Reasoning · Mathematical Reasoningaccuracy93.1%#6/22source ↗
02Tau2-BenchAgentic AI · Tool Useaccuracy80.3%#6/11source ↗
03LiveCodeBenchComputer Code · Code Generationpass-183.3%#8/24source ↗
04SWE-Bench VerifiedComputer Code · Code Generationaccuracy73.1%#12/22source ↗
05BrowseCompNatural Language Processing · Question Answeringaccuracy51.4%#13/16source ↗
06HLEReasoning · Multi-step Reasoningaccuracy25.1%#25/74source ↗
07GPQA DiamondReasoning · Multi-step Reasoningaccuracy82.4%#31/74source ↗
08PLCCNatural Language Processing · Polish Cultural Competencyculture-and-tradition78.0%#41/165source ↗
09PLCCNatural Language Processing · Polish Cultural Competencyhistory82.0%#53/165source ↗
10PLCCNatural Language Processing · Polish Cultural Competencyvocabulary65.0%#53/165source ↗
11PLCCNatural Language Processing · Polish Cultural Competencyaverage71.7%#54/165source ↗
12PLCCNatural Language Processing · Polish Cultural Competencyart-and-entertainment61.0%#61/165source ↗
13PLCCNatural Language Processing · Polish Cultural Competencygrammar66.0%#61/165source ↗
14PLCCNatural Language Processing · Polish Cultural Competencygeography78.0%#68/165source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area

Where DeepSeek-V3.2 actually performs.

Agentic AI
1
benchmark
avg rank #6.0
Computer Code
2
benchmarks
avg rank #10.0
Reasoning
3
benchmarks
avg rank #20.7
Natural Language Processing
2
benchmarks
avg rank #50.5
§ 04 · Papers

1 paper with results for DeepSeek-V3.2.

  1. 2025-12-02· 7 results

    DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models

§ 05 · Related models

Other DeepSeek models scored on Codesota.

DeepSeek-V4-Pro Max
4 results · 1 SOTA
DeepSeek R1
671B MoE params · 10 results
DeepSeek-V3
7 results
DeepSeek-Coder-V2-Instruct
Unknown params · 4 results
DeepSeek-OCR
4 results
DeepSeek-V4-Flash Max
4 results
DeepSeek-V3.2-Speciale
3 results
DeepSeek V3.5
685B MoE params · 2 results
§ 06 · Sources & freshness

Where these numbers come from.

pwc-dump
7
results
sdadas/PLCC
7
results
7 of 14 rows marked verified.