Gemma 4 E4B
Google
Code Multilingual Thinking Tool Calls Vision
Gemma 4 E4B is Google DeepMind's Effective 4B dense edge model, distilled from Gemini research for on-device and embedded deployment. It scores 69.4 on MMLU-Pro, 42.5 on AIME 2026, and 52.0 on LiveCodeBench v6, delivering strong reasoning in a compact footprint. Natively multimodal, it processes text, images, and audio with built-in thinking and tool-calling capabilities across a 128K context window. Released under the Apache 2.0 license, it requires only about 5 GB of VRAM at Q4, making it an excellent choice for self-hosted deployment on consumer GPUs and edge devices.
Hardware Configuration
Optional — for precise deployment recommendations
| Quantization | Quality | Size | Fit |
|---|---|---|---|
| FP16 | Full precision | 14.02 GB | — |
| BF16 | Full precision | 14.02 GB | — |
| Q8_0 | High | 7.48 GB | — |
| Q8_K_XL | High | 8.06 GB | — |
| Q6_K | High | 6.59 GB | — |
| Q6_K_XL | High | 6.95 GB | — |
| Q5_K_M | Medium | 5.11 GB | — |
| Q5_K_S | Medium | 5.03 GB | — |
| Q5_K_XL | Medium | 6.19 GB | — |
| Q4_K_M | Medium | 4.97 GB | — |
| Q4_K_S | Medium | 4.51 GB | — |
| Q4_K_XL | Medium | 4.75 GB | — |
| IQ4_NL | Medium | 4.5 GB | — |
| IQ4_XS | Medium | 4.39 GB | — |
| Q4_0 | Medium | 4.5 GB | — |
| Q4_1 | Medium | 4.73 GB | — |
| Q3_K_M | Low | 3.78 GB | — |
| Q3_K_S | Low | 3.6 GB | — |
| Q3_K_XL | Low | 4.25 GB | — |
| IQ3_XXS | Low | 3.45 GB | — |
| Q2_K_XL | Low | 3.49 GB | — |
| IQ2_M | Low | 3.29 GB | — |
Last updated: April 3, 2026