Provider Snapshot
10
48.16
0.54
Dec 26, 2025
Key Takeaways
10 openai models are actively benchmarked with 26358 total measurements across 25295 benchmark runs.
gpt-4.1-nano leads the fleet with 81.00 tokens/second, while GPT-5.1 delivers 39.10 tok/s.
Performance varies by 107.2% across the openai model lineup, indicating diverse optimization strategies for different use cases.
Median time to first token across the fleet is 0.54 ms, showing excellent responsiveness for interactive applications.
The openai model fleet shows varied performance characteristics (35.9% variation coefficient), reflecting diverse model architectures.
Fastest Models
| Provider | Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|---|
| openai | gpt-4.1-nano | 81.00 | 9.20 | 169.00 | 0.36 |
| openai | gpt-3.5-turbo | 75.50 | 1.36 | 137.00 | 0.53 |
| openai | gpt-4.1-mini | 56.90 | 9.62 | 129.00 | 0.33 |
| openai | gpt-4o | 53.40 | 3.69 | 128.00 | 0.58 |
| openai | gpt-4.1 | 41.50 | 6.91 | 94.60 | 0.38 |
| openai | GPT-5.1 | 39.10 | 7.31 | 76.80 | 0.77 |
All Models
Complete list of all openai models tracked in the benchmark system. Click any model name to view detailed performance data.
| Provider | Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|---|
| openai | gpt-4 | 25.60 | 4.71 | 49.20 | 0.83 |
| openai | gpt-4o | 53.40 | 3.69 | 128.00 | 0.58 |
| openai | gpt-4o-mini | 36.90 | 3.80 | 71.70 | 0.58 |
| openai | gpt-3.5-turbo | 75.50 | 1.36 | 137.00 | 0.53 |
| openai | gpt-4-turbo | 34.90 | 4.03 | 54.10 | 0.53 |
| openai | gpt-4.1 | 41.50 | 6.91 | 94.60 | 0.38 |
| openai | gpt-4.1-mini | 56.90 | 9.62 | 129.00 | 0.33 |
| openai | gpt-4.1-nano | 81.00 | 9.20 | 169.00 | 0.36 |
| openai | GPT-5.1 | 39.10 | 7.31 | 76.80 | 0.77 |
| openai | GPT-5.2 | 36.80 | 12.90 | 58.90 | 0.53 |
Featured Models
Frequently Asked Questions
Based on recent tests, gpt-4.1-nano shows the highest average throughput among tracked openai models.
This provider summary aggregates 26358 individual prompts measured across 25295 monitoring runs over the past month.