Compare LLM Models
Select 2-5 models to compare side by side.
Select models to compare
0 / 5 selected
Top downloads
Select 2-5 models
| Attribute |
allenai
|
|---|---|
| Company | allenai |
| Release Date | 2026-03-05 |
| Parameters (B) | 7 |
| Architecture | Hybrid RNN-Transformer (Gated DeltaNet + Attention) |
| Context Window | 65536 |
| Input Modalities | text |
| Open Source | Yes |
| License | apache-2.0 |
| Score | 0 |
| VRAM (GB) | 14 |
| Compute | Single NVIDIA GPU with 16GB+ VRAM for BF16 inference; trained on 512 H100/B200 GPUs |
| Benchmark: ARC | 90.8 |
| Benchmark: BBH | 65.2 |
| Benchmark: DROP | 72.9 |
| Benchmark: HellaSwag | 79 |
| Benchmark: HumanEval | 49 |
| Benchmark: MATH | 23.4 |
| Benchmark: MBPP | 50.3 |
| Benchmark: MMLU-Pro | 41.7 |
| Pricing (Input) | — |
| Pricing (Output) | — |