DeepSeek V3.1 NEX N1: benchmark results
Nex AGI's open-weight DeepSeek-V3.1-Nex-N1 (December 2025), the flagship 671B-parameter release of the Nex-N1 series, a post-trained fine-tune of DeepSeek-V3.1 built for agent autonomy and tool use. Provider: Nex AGI. Released 2025-12-08. Access: Open.
Aggregate Rating 1640 ± 25, rank #335 of 1635 rated models, from 80 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AGC-Bench - ocw_connections | 1.1 | Dataset z-score | 95.1 |
| AGC-Bench - poetmt | 1.25 | Dataset z-score | 91.4 |
| AGC-Bench - cpers | 1.05 | Dataset z-score | 90.2 |
| AGC-Bench - scar | 0.86 | Dataset z-score | 90.2 |
| AGC-Bench - schnovel | 1.2 | Dataset z-score | 89.6 |
| AGC-Bench - STEM | 0.54 | JRT z-score | 87.8 |
| AGC-Bench - grapheval_review_advisor | 1.06 | Dataset z-score | 87.8 |
| AGC-Bench - conceptual_design | 0.96 | Dataset z-score | 85.4 |
| AGC-Bench - tinyfabulist | 0.92 | Dataset z-score | 85.4 |
| AGC-Bench - slang_generation | 0.92 | Dataset z-score | 84.1 |
| AGC-Bench - rpgbench | 1.06 | Dataset z-score | 84 |
| AGC-Bench - brainteaser | 0.86 | Dataset z-score | 83.5 |
Interactive version: theaggregate.ai/model?slug=deepseek-v3-1-nex-n1 · How It Works · Data refreshed daily, snapshot 2026-11-05.