Provider Snapshot
31
60.24
0.56
Dec 24, 2025
Key Takeaways
31 bedrock models are actively benchmarked with 34063 total measurements across 30805 benchmark runs.
nova-micro leads the fleet with 135.00 tokens/second, while nova-lite delivers 98.40 tok/s.
Performance varies by 37.2% across the bedrock model lineup, indicating diverse optimization strategies for different use cases.
Median time to first token across the fleet is 0.56 ms, showing excellent responsiveness for interactive applications.
The bedrock model fleet shows varied performance characteristics (54.0% variation coefficient), reflecting diverse model architectures.
Fastest Models
| Provider | Model | Avg Toks/Sec | Min | Max | Avg TTF (ms) |
|---|---|---|---|---|---|
| bedrock | nova-micro | 135.00 | 83.20 | 170.00 | 0.21 |
| bedrock | llama-4-maverick | 116.00 | 39.30 | 149.00 | 0.23 |
| bedrock | llama-4-scout | 107.00 | 32.30 | 140.00 | 0.24 |
| bedrock | llama-3.1-8b | 103.00 | 40.40 | 120.00 | 0.32 |
| bedrock | llama-3.3-70b | 103.00 | 8.55 | 142.00 | 0.21 |
| bedrock | nova-lite | 98.40 | 2.10 | 138.00 | 0.28 |
All Models
Complete list of all bedrock 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) |
|---|---|---|---|---|---|
| bedrock | claude-3-haiku | 64.00 | 14.90 | 91.40 | 0.27 |
| bedrock | claude-3-5-haiku | 31.60 | 2.78 | 39.00 | 0.68 |
| bedrock | claude-3-sonnet | 43.20 | 4.49 | 58.70 | 0.29 |
| bedrock | claude-3-5-sonnet | 30.10 | 7.05 | 46.00 | 0.64 |
| bedrock | claude-3-7-sonnet | 31.20 | 4.64 | 42.50 | 0.81 |
| bedrock | claude-3-opus | 18.40 | 8.52 | 21.90 | 0.86 |
| bedrock | llama-3-8b | 81.70 | 7.03 | 89.60 | 0.20 |
| bedrock | llama-3-70b | 40.20 | 7.43 | 43.30 | 0.23 |
| bedrock | llama-3.1-8b | 103.00 | 40.40 | 120.00 | 0.32 |
| bedrock | llama-3.1-70b | 25.10 | 12.80 | 26.20 | 0.56 |
| bedrock | llama-3.1-405b | 18.20 | 9.23 | 20.30 | 1.42 |
| bedrock | llama-3.2-1b | 65.80 | 14.80 | 72.30 | 0.26 |
| bedrock | llama-3.2-3b | 43.90 | 3.24 | 48.50 | 0.30 |
| bedrock | llama-3.2-11b | 91.10 | 1.03 | 119.00 | 0.50 |
| bedrock | llama-3.2-90b | 48.10 | 17.00 | 51.20 | 0.33 |
| bedrock | mistral-7b | 85.90 | 3.26 | 93.10 | 0.17 |
| bedrock | mixtral-8x7b | 82.20 | 7.09 | 90.60 | 0.18 |
| bedrock | mistral-small | 59.50 | 12.40 | 62.30 | 0.16 |
| bedrock | mistral-large | 44.90 | 9.17 | 48.10 | 0.22 |
| bedrock | command-r | 74.70 | 41.10 | 79.10 | 0.14 |
| bedrock | command-r-plus | 36.00 | 4.39 | 37.40 | 0.17 |
| bedrock | nova-pro | 75.40 | 5.77 | 121.00 | 0.37 |
| bedrock | nova-lite | 98.40 | 2.10 | 138.00 | 0.28 |
| bedrock | nova-micro | 135.00 | 83.20 | 170.00 | 0.21 |
| bedrock | llama-3.3-70b | 103.00 | 8.55 | 142.00 | 0.21 |
| bedrock | llama-4-maverick | 116.00 | 39.30 | 149.00 | 0.23 |
| bedrock | llama-4-scout | 107.00 | 32.30 | 140.00 | 0.24 |
| bedrock | claude-sonnet-4.5 | 20.30 | 2.02 | 26.70 | 1.99 |
| bedrock | titan-text-lite | 26.50 | 20.70 | 31.00 | 2.43 |
| bedrock | claude-opus-4.5 | 19.60 | 1.11 | 23.40 | 2.02 |
| bedrock | claude-haiku-4.5 | 47.50 | 4.47 | 65.40 | 0.76 |
Featured Models
Frequently Asked Questions
Based on recent tests, nova-micro shows the highest average throughput among tracked bedrock models.
This provider summary aggregates 34063 individual prompts measured across 30805 monitoring runs over the past month.