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 |
Qwen
|
|---|---|
| Company | Qwen |
| Release Date | 2024-11-06 |
| Parameters (B) | 14.7 |
| Architecture | Transformer |
| Context Window | 131072 |
| Input Modalities | text |
| Open Source | Yes |
| License | apache-2.0 |
| Score | 0 |
| VRAM (GB) | 28 |
| Compute | ~28GB VRAM for BF16 inference; ~10GB with Q4 quantization. Single NVIDIA GPU (RTX 4090/3090 or better recommended). 48-layer Transformer with GQA (40Q/8KV heads). |
| Benchmark: AIME-2024 | 10 |
| Benchmark: Aider | 58.6 |
| Benchmark: C-Eval | 66.2 |
| Benchmark: GPQA | 36.8 |
| Benchmark: GSM8K | 94.2 |
| Benchmark: HumanEval | 70.7 |
| Benchmark: IFEval | 66.5 |
| Benchmark: LiveCodeBench | 23.4 |
| Benchmark: MATH | 66.8 |
| Benchmark: MBPP | 69.2 |
| Benchmark: MMLU | 71.7 |
| Benchmark: MMLU-Pro | 55.6 |
| Pricing (Input) | — |
| Pricing (Output) | — |