8 builds of Gemma 4 E4B measured on real machines — different packagers, precisions and engines, each a different file with its own numbers. The best each engine proved is first; every build has its own page with the machines and signatures behind it. 8B parameters, as the vendor states.
trained 128K · proved 128K context proved by google Q4_K_M
93.7 tok/s peak, by google Q4_K_M
proved by at least one build: Tool calling · Tool loop finished · Faithful to tool results · Chained tool calls · Structured output · Sustained output · Extended thinking · Vision · Counts in images · Multiple images at once · Spatial reasoning · Writes code · Reasons about code · Synthesises sources · Follows instructions
| build | publisher | precision | engine | whose work | context proved | best tok/s | machines |
|---|---|---|---|---|---|---|---|
| gemma4:e4b | Q4_K_M | ollama | vendor | 1,31,072 | 93.7 | 1 | |
| hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ4_XS | unsloth | IQ4_XS | ollama | quantizer | 1,31,072 | 61.4 | 1 |
| hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_K_M | unsloth | Q4_K_M | ollama | quantizer | 1,31,072 | 56.8 | 1 |
| hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_0 | unsloth | Q4_0 | ollama | quantizer | 1,31,072 | 54.7 | 1 |
| hf.co/unsloth/gemma-4-E4B-it-GGUF:Q8_0 | unsloth | Q8_0 | ollama | quantizer | 1,31,072 | 39.5 | 1 |
| hf.co/unsloth/gemma-4-E4B-it-GGUF:BF16 | unsloth | BF16 | ollama | quantizer | 1,31,072 | 25.8 | 1 |
| hf.co/unsloth/gemma-4-E4B-it-GGUF:Q3_K_M | unsloth | Q3_K_M | ollama | quantizer | 4,096 | 20.4 | 1 |
| gemma4:e4b-it-qat | ollama | — | ollama | — | — | — | 1 |