Provider Snapshot
16
36.58
1385.63
Apr 3, 2026
Key Takeaways
16 deepinfra models are actively benchmarked with 3746 total measurements across 3648 benchmark runs.
mistral-7b leads the fleet with 89.10 tokens/second, while llama-3.2-3b delivers 40.10 tok/s.
Performance varies by 122.2% across the deepinfra model lineup, indicating diverse optimization strategies for different use cases.
Avg time to first token across the fleet is 1385.63 ms, showing moderate responsiveness for interactive applications.
The deepinfra model fleet shows varied performance characteristics (55.6% variation coefficient), reflecting diverse model architectures.
Fastest Models
| Provider | Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|---|
| deepinfra | mistral-7b | 89.10 | 10.80 | 148.00 | 490.00 |
| deepinfra | devstral-small | 77.30 | 9.24 | 140.00 | 540.00 |
| deepinfra | Qwen 2.5 Coder 32B | 45.90 | 1.47 | 83.50 | 2120.00 |
| deepinfra | llama-3-8b | 45.30 | 18.80 | 68.90 | 320.00 |
| deepinfra | llama-3.2-1b | 40.80 | 3.15 | 100.00 | 750.00 |
| deepinfra | llama-3.2-3b | 40.10 | 3.02 | 99.00 | 750.00 |
All Models
Complete list of all deepinfra 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) |
|---|---|---|---|---|---|
| deepinfra | Qwen 2.5 Coder 32B | 45.90 | 1.47 | 83.50 | 2120.00 |
| deepinfra | devstral-small | 77.30 | 9.24 | 140.00 | 540.00 |
| deepinfra | llama-3.3-70b | 18.70 | 1.03 | 42.90 | 2210.00 |
| deepinfra | qwen-2.5-72b | 28.60 | 1.07 | 45.90 | 2470.00 |
| deepinfra | mistral-7b | 89.10 | 10.80 | 148.00 | 490.00 |
| deepinfra | llama-3.2-11b | 24.60 | 1.81 | 80.50 | 1470.00 |
| deepinfra | llama-3.2-3b | 40.10 | 3.02 | 99.00 | 750.00 |
| deepinfra | llama-3.2-1b | 40.80 | 3.15 | 100.00 | 750.00 |
| deepinfra | llama-3.1-405b | 21.80 | 1.35 | 38.80 | 1120.00 |
| deepinfra | llama-3.1-70b | 20.80 | 1.02 | 41.30 | 1080.00 |
| deepinfra | llama-3.1-8b | 35.60 | 2.42 | 78.00 | 760.00 |
| deepinfra | llama-3-70b | 29.10 | 4.22 | 50.50 | 570.00 |
| deepinfra | llama-3-8b | 45.30 | 18.80 | 68.90 | 320.00 |
| deepinfra | llama-2-70b | 29.00 | 4.28 | 51.30 | 630.00 |
| deepinfra | qwen-3-235b | 6.91 | 1.07 | 52.60 | 6080.00 |
| deepinfra | llama-3.2-90b | 31.60 | 4.35 | 81.30 | 810.00 |
Featured Models
Frequently Asked Questions
Based on recent tests, mistral-7b shows the highest average throughput among tracked deepinfra models.
This provider summary aggregates 3746 individual prompts measured across 3648 monitoring runs over the past month.