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 |
LiquidAI
|
|---|---|
| Company | LiquidAI |
| Release Date | 2025-10-07 |
| Parameters (B) | 8.3 |
| Architecture | LFM2 Hybrid MoE |
| Context Window | 32768 |
| Input Modalities | text |
| Open Source | Yes |
| License | other |
| Score | 1 |
| VRAM (GB) | 16 |
| Compute | ~16GB VRAM for BF16 inference (8.3B total MoE weights); INT4 quantization runs on consumer laptops, tablets, and high-end phones |
| Benchmark: GPQA | 29.3 |
| Benchmark: GSM8K | 84.4 |
| Benchmark: HumanEval | 69.5 |
| Benchmark: IFEval | 77.6 |
| Benchmark: LiveCodeBench | 21.0 |
| Benchmark: MATH-500 | 74.2 |
| Benchmark: MMLU | 64.8 |
| Benchmark: MMLU-Pro | 37.4 |
| Pricing (Input) | — |
| Pricing (Output) | — |