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 |
OpenGVLab
|
|---|---|
| Company | OpenGVLab |
| Release Date | 2025-04-11 |
| Parameters (B) | 2.1 |
| Architecture | ViT-MLP-LLM |
| Context Window | 32768 |
| Input Modalities | text, image |
| Open Source | Yes |
| License | apache-2.0 |
| Score | 0 |
| VRAM (GB) | 8 |
| Compute | Single NVIDIA GPU with 8GB+ VRAM for bf16 inference (4.2GB model weights); 5GB+ with 8-bit quantization. Flash Attention recommended; supports multi-GPU sharding via device_map. |
| Benchmark: BBH | 52 |
| Benchmark: C-Eval | 73.3 |
| Benchmark: GSM8K | 72.5 |
| Benchmark: HellaSwag | 73.8 |
| Benchmark: HumanEval | 62.8 |
| Benchmark: MATH | 57.3 |
| Benchmark: MBPP | 60.7 |
| Benchmark: MMLU | 64.8 |
| Benchmark: MMMU | 48.6 |
| Pricing (Input) | — |
| Pricing (Output) | — |