GPT-4.1 nano
Multimodal
Zero-eval
#3OpenAI-MRCR: 2 needle 1M
#3Graphwalks parents >128k
#3Graphwalks BFS >128k
by OpenAI
+
+
+
+
About
GPT-4.1 Nano was developed as the smallest and most efficient variant in the GPT-4.1 family, designed for applications requiring minimal latency and resource usage. Built to enable AI capabilities on edge devices and resource-constrained environments, it distills GPT-4.1 capabilities into an ultra-compact form factor.
+
+
+
+
Pricing Range
Input (per 1M)$0.10 -$0.10
Output (per 1M)$0.40 -$0.40
Providers1
+
+
+
+
Timeline
AnnouncedApr 14, 2025
ReleasedApr 14, 2025
Knowledge CutoffMay 31, 2024
+
+
+
+
Specifications
Capabilities
Multimodal
+
+
+
+
License & Family
License
Proprietary
Performance Overview
Performance metrics and category breakdown
Overall Performance
25 benchmarks
Average Score
34.2%
Best Score
80.1%
High Performers (80%+)
1Performance Metrics
Max Context Window
1.1MAvg Throughput
200.0 tok/sAvg Latency
2ms+
+
+
+
All Benchmark Results for GPT-4.1 nano
Complete list of benchmark scores with detailed information
| MMLU | text | 0.80 | 80.1% | Self-reported | |
| IFEval | text | 0.74 | 74.5% | Self-reported | |
| CharXiv-D | multimodal | 0.74 | 73.9% | Self-reported | |
| MMMLU | text | 0.67 | 66.9% | Self-reported | |
| Multi-IF | text | 0.57 | 57.2% | Self-reported | |
| MathVista | multimodal | 0.56 | 56.2% | Self-reported | |
| MMMU | multimodal | 0.55 | 55.4% | Self-reported | |
| GPQA | text | 0.50 | 50.3% | Self-reported | |
| COLLIE | text | 0.42 | 42.5% | Self-reported | |
| CharXiv-R | multimodal | 0.41 | 40.5% | Self-reported |
Showing 1 to 10 of 25 benchmarks
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+