Provider Snapshot
46
43.78
0.81
Dec 24, 2025
Key Takeaways
46 openrouter models are actively benchmarked with 44111 total measurements across 43144 benchmark runs.
llama-3.2-1b leads the fleet with 111.00 tokens/second, while codestral delivers 70.10 tok/s.
Performance varies by 58.3% across the openrouter model lineup, indicating diverse optimization strategies for different use cases.
Median time to first token across the fleet is 0.81 ms, showing excellent responsiveness for interactive applications.
The openrouter model fleet shows varied performance characteristics (42.9% variation coefficient), reflecting diverse model architectures.
Fastest Models
| Provider | Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|---|
| openrouter | llama-3.2-1b | 111.00 | 28.00 | 159.00 | 0.37 |
| openrouter | mistral-7b | 86.10 | 1.70 | 156.00 | 0.30 |
| openrouter | ministral-3b | 83.30 | 5.66 | 114.00 | 0.37 |
| openrouter | mistral-saba | 80.30 | 6.79 | 124.00 | 0.39 |
| openrouter | ministral-8b | 75.00 | 4.84 | 89.10 | 0.38 |
| openrouter | codestral | 70.10 | 22.10 | 97.10 | 0.49 |
All Models
Complete list of all openrouter 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) |
|---|---|---|---|---|---|
| openrouter | qwen-3-30b | 42.60 | 5.11 | 124.00 | 0.99 |
| openrouter | qwen-3-8b | 46.00 | 1.15 | 135.00 | 1.40 |
| openrouter | qwen-3-14b | 32.50 | 1.41 | 77.20 | 1.32 |
| openrouter | qwen-3-32b | 34.20 | 1.96 | 167.00 | 1.03 |
| openrouter | qwen-3-235b | 36.00 | 1.30 | 91.30 | 0.76 |
| openrouter | gpt-4.1 | 36.70 | 9.18 | 80.60 | 0.21 |
| openrouter | gpt-4.1-mini | 39.30 | 12.60 | 85.60 | 0.21 |
| openrouter | gpt-4.1-nano | 46.20 | 1.65 | 103.00 | 0.31 |
| openrouter | llama-3.1-nemotron-ultra | 35.10 | 8.69 | 38.10 | 0.33 |
| openrouter | qwen-2.5-vl | 26.80 | 1.07 | 68.90 | 1.03 |
| openrouter | mistral-small | 35.10 | 3.12 | 118.00 | 1.09 |
| openrouter | qwq-32b | 13.50 | 2.66 | 17.50 | 3.02 |
| openrouter | mistral-saba | 80.30 | 6.79 | 124.00 | 0.39 |
| openrouter | qwen-vl-plus | 45.60 | 11.30 | 71.10 | 0.82 |
| openrouter | qwen-vl-max | 32.60 | 1.46 | 45.00 | 0.90 |
| openrouter | deepseek-r1 | 38.90 | 2.10 | 99.30 | 1.04 |
| openrouter | mistral-large | 41.50 | 1.44 | 54.60 | 0.73 |
| openrouter | pixtral-large | 36.70 | 1.44 | 55.80 | 0.70 |
| openrouter | qwen-2.5-coder | 32.10 | 3.53 | 73.50 | 0.93 |
| openrouter | qwen-2.5-7b | 47.40 | 1.24 | 128.00 | 1.09 |
| openrouter | llama-3.1-nemotron-70b | 40.40 | 3.12 | 47.10 | 0.40 |
| openrouter | llama-3.2-3b | 45.40 | 5.45 | 133.00 | 0.49 |
| openrouter | llama-3.2-1b | 111.00 | 28.00 | 159.00 | 0.37 |
| openrouter | llama-3.2-11b | 28.50 | 1.93 | 104.00 | 1.17 |
| openrouter | llama-3.2-90b | 42.50 | 3.09 | 87.00 | 0.41 |
| openrouter | qwen-2.5-72b | 33.80 | 2.49 | 72.00 | 0.74 |
| openrouter | llama-3.1-8b | 50.10 | 4.34 | 271.00 | 0.65 |
| openrouter | qwen-2.5-vl-7b | 57.80 | 1.57 | 81.70 | 0.64 |
| openrouter | llama-3.1-70b | 32.40 | 1.35 | 127.00 | 0.76 |
| openrouter | llama-3.1-405b | 18.70 | 1.12 | 30.10 | 1.36 |
| openrouter | gpt-4o | 52.20 | 3.50 | 112.00 | 0.53 |
| openrouter | mistral-nemo | 34.90 | 7.48 | 121.00 | 0.79 |
| openrouter | gpt-4o-mini | 39.40 | 10.90 | 87.60 | 0.51 |
| openrouter | mistral-7b | 86.10 | 1.70 | 156.00 | 0.30 |
| openrouter | llama-3.3-70b | 34.20 | 1.45 | 234.00 | 0.77 |
| openrouter | devstral-small | 39.70 | 2.28 | 100.00 | 0.41 |
| openrouter | mistral-medium | 28.60 | 1.08 | 50.30 | 1.60 |
| openrouter | pixtral-12b | 57.30 | 2.37 | 74.20 | 0.55 |
| openrouter | deepseek-chat | 26.90 | 10.50 | 51.80 | 1.38 |
| openrouter | codestral | 70.10 | 22.10 | 97.10 | 0.49 |
| openrouter | devstral-medium | 39.40 | 4.44 | 49.80 | 0.89 |
| openrouter | deepseek-v3 | 19.20 | 2.68 | 111.00 | 1.33 |
| openrouter | magistral-medium | 30.60 | 2.39 | 39.60 | 1.24 |
| openrouter | magistral-small | 57.50 | 16.90 | 72.10 | 0.49 |
| openrouter | ministral-8b | 75.00 | 4.84 | 89.10 | 0.38 |
| openrouter | ministral-3b | 83.30 | 5.66 | 114.00 | 0.37 |
Featured Models
Frequently Asked Questions
Based on recent tests, llama-3.2-1b shows the highest average throughput among tracked openrouter models.
This provider summary aggregates 44111 individual prompts measured across 43144 monitoring runs over the past month.