RUN THIS LLM
Search local LLM hardware requirements
Gemma 3 4B
Google · 4B · Vision
Compact multimodal model with vision support. Strong efficiency for its size class.
VRAM Requirements
| Quantization | VRAM |
|---|---|
| Q4_K_M (smallest) | 2.4 GB |
| Q8_0 (balanced) | 4.4 GB |
| FP16 (full quality) | 8 GB |
Specifications
- Parameters: 4B
- Category: Vision
- Max context: 128K tokens
- System RAM: 6 GB minimum
- HuggingFace: google/gemma-3-4b-it
Benchmarks
- MMLU: 64 — General knowledge & reasoning (5-shot)
- HumanEval: 48 — Code generation (pass@1)
- MATH: 32 — Competition-level math reasoning
Loading interactive analysis...