DeepSeek V3.2 Speciale: benchmark results
DeepSeek V3.2 Speciale variant row. Provider: DeepSeek. Released 2025-12-01. Access: Open.
Unified ELO 1656 ± 1, rank #119 of 1392 rated models, from 113 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MathArena - CMIMC 2025 | 94.38 | Accuracy (%) | 100 |
| OpenCompass Research - AIME 2025 | 96 | Score (%) | 100 |
| UGI Leaderboard | 65.37 | UGI Score | 99.5 |
| JudgeBench Reasoning | 96.94 | Accuracy (%) | 98 |
| OpenCompass Research - HLE | 28.6 | Score (%) | 96.8 |
| BenchTable | 76.9 | Total Score (%) | 94.5 |
| LLM Stats (HMMT 2025) | 99.2 | Score (%) | 93.8 |
| MathArena - HMMT Feb 2025 | 97.5 | Accuracy (%) | 92.6 |
| MathArena - BRUMO 2025 | 99.17 | Accuracy (%) | 92.2 |
| OpenCompass Research - GPQA Diamond | 86.7 | Score (%) | 91.5 |
| UGI - Natural Intelligence | 53.02 | NatInt Score | 91 |
| OpenCompass | 78.2 | Average Score | 90.4 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-2-speciale · How It Works · Data refreshed daily, snapshot 2026-09-05.