DeepSeek V3.2 Exp (Thinking) — benchmark results
DeepSeek V3.2 Exp evaluated with thinking enabled. Provider: DeepSeek. Released 2025-09-29. Access: Open.
Unified ELO 1692 ± 21, rank #286 of 1776 rated models, from 70 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| UGI Leaderboard | 53.84 | UGI Score | 95.5 |
| AA LiveCodeBench | 78.94 | Pass@1 (%) | 91.1 |
| AA Long Context Reasoning | 69 | Accuracy (%) | 91 |
| UGI - Writing | 54 | Writing Score | 91 |
| AA MMLU-Pro | 85.04 | Accuracy (%) | 90.8 |
| UGI - Natural Intelligence | 46.34 | NatInt Score | 89.4 |
| Gorilla API Bench (BFCL) | 56.73 | Overall Accuracy (%) | 86.6 |
| AA AIME 2025 | 87.67 | Accuracy (%) | 85 |
| AA Global-MMLU-Lite - German | 91.08 | Accuracy (%) | 83.9 |
| AA Global-MMLU-Lite - French | 91.08 | Accuracy (%) | 82.9 |
| AA Global-MMLU-Lite - Portuguese | 91 | Accuracy (%) | 81.7 |
| AA Omniscience - Software Engineering (SWE) - Swift | 60 | Accuracy (%) | 80.9 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-2-exp-thinking · How the rankings work · Data refreshed daily, snapshot 2026-07-22.