Provider Snapshot
17
31.58
1080.00
Feb 8, 2026
Key Takeaways
17 deepinfra models are actively benchmarked with 3972 total measurements across 3957 benchmark runs.
mixtral-8x22b leads the fleet with 49.40 tokens/second, while llama-3.1-8b delivers 36.30 tok/s.
Performance varies by 36.1% across the deepinfra model lineup, indicating diverse optimization strategies for different use cases.
Avg time to first token across the fleet is 1080.00 ms, showing moderate responsiveness for interactive applications.
The deepinfra model fleet shows varied performance characteristics (31.9% variation coefficient), reflecting diverse model architectures.
Fastest Models
| Provider | Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|---|
| deepinfra | mixtral-8x22b | 49.40 | 26.20 | 79.40 | 300.00 |
| deepinfra | llama-3-8b | 45.50 | 7.50 | 70.40 | 300.00 |
| deepinfra | mistral-7b | 45.30 | 3.28 | 89.00 | 540.00 |
| deepinfra | devstral-small | 42.10 | 3.02 | 84.80 | 500.00 |
| deepinfra | llama-3.2-90b | 38.20 | 1.77 | 92.70 | 860.00 |
| deepinfra | llama-3.1-8b | 36.30 | 1.26 | 102.00 | 1250.00 |
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 | 31.80 | 1.01 | 67.00 | 4670.00 |
| deepinfra | devstral-small | 42.10 | 3.02 | 84.80 | 500.00 |
| deepinfra | llama-3.3-70b | 23.40 | 1.45 | 58.40 | 820.00 |
| deepinfra | qwen-3-235b | 22.10 | 1.44 | 42.50 | 530.00 |
| deepinfra | qwen-2.5-72b | 31.80 | 2.68 | 46.20 | 900.00 |
| deepinfra | mistral-7b | 45.30 | 3.28 | 89.00 | 540.00 |
| deepinfra | mixtral-8x22b | 49.40 | 26.20 | 79.40 | 300.00 |
| deepinfra | llama-3.2-90b | 38.20 | 1.77 | 92.70 | 860.00 |
| deepinfra | llama-3.2-11b | 13.90 | 1.59 | 67.80 | 2040.00 |
| deepinfra | llama-3.2-3b | 26.60 | 1.03 | 89.20 | 650.00 |
| deepinfra | llama-3.2-1b | 26.20 | 4.56 | 82.20 | 380.00 |
| deepinfra | llama-3.1-405b | 16.90 | 1.05 | 30.90 | 3190.00 |
| deepinfra | llama-3.1-70b | 21.30 | 3.70 | 38.70 | 530.00 |
| deepinfra | llama-3.1-8b | 36.30 | 1.26 | 102.00 | 1250.00 |
| deepinfra | llama-3-70b | 32.80 | 2.21 | 44.00 | 520.00 |
| deepinfra | llama-3-8b | 45.50 | 7.50 | 70.40 | 300.00 |
| deepinfra | llama-2-70b | 33.30 | 8.00 | 43.50 | 380.00 |
Featured Models
Frequently Asked Questions
Based on recent tests, mixtral-8x22b shows the highest average throughput among tracked deepinfra models.
This provider summary aggregates 3972 individual prompts measured across 3957 monitoring runs over the past month.