NVIDIANemotron 3 Super 120B A12B (Reasoning)
- Released
- Mar 11, 2026
- Tracked since
- 2026-07-23
Intelligence index12.8Ranks 310 of 664 models with this metric
Coding index37.7Ranks 147 of 259 models with this metric
Agentic index1.7Ranks 121 of 156 models with this metric
Capability profile
5 axes vs. site averageNemotron 3 Super 120B A12B (Reasoning)
Site average
Pricing
per 1M tokens- Input
- $0.300
- Output
- $0.900
- Cache read
- $0.150
Performance
- Output throughput
- 169 tokens/sec
- Time to first token
- 1.45s
- First answer token
- 13.26s
- End-to-end response
- 16.21s
History
12Intelligence index
25.4→12.8
Output
$0.780→$0.900
Output throughput
209→169
Cheaper at this level
Models with a comparable intelligence index and a lower output price- Qwen3.5 4B (Reasoning)
AlibabaIntelligence 13.1$0.150−83% - DeepSeek V4 Flash 0420 (Non-reasoning)
DeepSeekIntelligence 18.9$0.200−78% - Qwen3.5 9B (Reasoning)AlibabaIntelligence 13.7$0.200−78%
- Hy3-preview (Reasoning)
TencentIntelligence 22.7$0.210−77% - Hy3-preview (Non-reasoning)TencentIntelligence 17$0.210−77%