evals.report
BenchmarksLabsCompareRun guides
DeepSeekDeepSeek

DeepSeek V3.2

DeepSeek · DeepSeek. Released Sep 29, 2025.

4 results

Benchmark results 4

Compare this model
BenchmarkCategoryScoreMetricStatusDate
FrontierMathReasoning22.1%accuracyOfficialMay 30, 2026Details
Berkeley Function Calling LeaderboardTool use56.73%accuracyOfficialApr 12, 2026Details
SWE-bench ProCoding15.56%% resolvedOfficialMay 30, 2026Details
AIME (OTIS Mock)Reasoning87.8%accuracyOfficialMay 30, 2026Details