Context
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| End-to-end Attention Latency | 120K context | Attention Latency (120K Context, µs)62.9 | 20 | |
| Open-Vocabulary Semantic Segmentation | Context59 | mIoU49.4 | 11 | |
| Prefill Throughput | 128K context | Prefill Throughput (tok/s)7,756.8 | 10 | |
| Prefill Throughput | 96K context | Prefill Throughput (tok/s)7,959.4 | 10 | |
| Prefill Throughput | 32K context | Prefill Throughput (tok/s)9,576.2 | 10 | |
| Inference efficiency | 128K-context | TTFT101 | 8 |
Showing 6 of 6 rows