o1 mini
by OpenAI
+
+
+
+
About
OpenAI o1 mini, released in September 2024, was designed as a cost-efficient reasoning model for STEM applications — applying the o1 chain-of-thought training approach at a smaller scale optimized for math, coding, and science problems. It gave developers access to reasoning model capabilities at significantly lower cost than full o1, making RL-trained reasoning practical for higher-volume applications.
+
+
+
+
Pricing Range
Input (per 1M)$1.10 -$1.10
Output (per 1M)$4.40 -$4.40
Providers1
+
+
+
+
Timeline
ReleasedSep 12, 2024
+
+
+
+
License & Family
License
Proprietary
Performance Overview
Performance metrics and category breakdown
Overall Performance
1 benchmarks
Average Score
0.9%
Best Score
0.9%
High Performers (80%+)
0Performance Metrics
Max Context Window
193.5KTop Categories
Agents
0.9%
+
+
+
+
All Benchmark Results for o1 mini
Complete list of benchmark scores with detailed information
| BrowseComp | Agents | 0.90 | 0.9% | Unverified |