family member
Gemma 4 logo

Gemma 4 E4B

83 tokens a second on Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it.

8 builds of Gemma 4 E4B, 8 signed runs on real machines: different packagers, precisions and engines, each a different file with its own numbers. 8B parameters, as the vendor states.

83tok/s, fastest machine
$1.73per M tokens · no API sells it
128Kcontext proved by recall
3.6 GBmemory, smallest build
3providers measured it
Use it in your own room Ask Evo in your gramx and it runs on a machine that proved it, at the metered $1.73 per million tokens. On your own machines it costs you nothing.Measure it on your machine A gram works through the catalog by itself and signs every result. 3 providers have measured Gemma 4 E4B; every run is a point on the providers board.Compare every build Put it beside every other model in one table: speed, price, context and the API price, one row each.
best on ollama · 8 builds

trained 128K · proved 128K context, by google Q4_K_M

93.7 tok/s peak, by google Q4_K_M

Speed against price, every machine

Up and left is faster and cheaper · the line joins machines no other beats on both · price on a log scale

USD per million tokenstok/sLinux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM · 4-bit by google · 82.83 tok/s · $1.7329/MLinux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM · 8-bit by unsloth · 38.18 tok/s · $3.6915/MLinux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM · 16-bit (full precision) by unsloth · 25.71 tok/s · $5.6897/M

What it fits on

The smallest build needs about 3.6 GB; 13 of 13 kinds of machine on the network can hold it. Bars: each kind's memory (its card's, else RAM) · ticks: every build, at the size it held plus a fifth for the prompt.

  1. NVIDIA GeForce RTX 3050 Laptop GPU4 GB on the card1 of 6
  2. NVIDIA GeForce GTX 16504 GB on the card1 of 6
  3. AMD Radeon(TM) Vega 8 Mobile Graphics7 GB RAM4 of 6
  4. Windows, CPU only · 12 CPUs7 GB RAM4 of 6
  5. apple8 GB RAM4 of 6
  6. Apple M18 GB RAM4 of 6
  7. NVIDIA GeForce RTX 3080 Ti12 GB on the card5 of 6
  8. Tesla T415 GB on the cardevery build
  9. Intel(R) UHD Graphics 73016 GB RAMevery build
  10. Intel(R) Iris(R) Xe Graphics16 GB RAMevery build
  11. Apple M416 GB RAMevery build
  12. 2× Tesla T429 GB on the cardevery build
  13. Linux, CPU only · 4 CPUs31 GB RAMevery build

How far back it remembers, machine by machine

Filled: the needle came back at that prompt size · outlined: asked and lost · blank: not asked · trained for 128K. 0 walks ended because the model forgot, 0 because the machine could not hold the prompt.

machine · build4K8K16K32K64K128Kwhy it stopped
Gram ·a5be46Tesla T4 · Q4_K_M by google128K recalled at every size asked
Gram ·a5be46Tesla T4 · IQ4_XS by unsloth128K recalled at every size asked
Gram ·a5be46Tesla T4 · Q4_K_M by unsloth128K recalled at every size asked
Gram ·a5be46Tesla T4 · Q4_0 by unsloth128K recalled at every size asked
Gram ·a5be46Tesla T4 · Q8_0 by unsloth128K recalled at every size asked
Gram ·a5be46Tesla T4 · BF16 by unsloth128K recalled at every size asked
nitin singhApple M4 · Q3_K_M by unsloth4K
Amit KhetanIntel(R) UHD Graphics 730 · by ollama—

Carbon per million tokens

Lowest build on each machine · 1 of 1 read off a power sensor, the rest modelled from the parts · fewer grams is better

  1. Gram ·a5be46353 g2.69 J/tok · sensor

Carbon is location-based: signed energy × the yearly average of the grid where the machine signed its place (world grid average (place not signed); Ember Global Electricity Review 2025 (2024 generation), location-based). Facility overhead is not included. A machine on its own solar is not credited until it can prove it.

Measured by

The people whose machines signed these numbers

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

Every build

Each its own file with its own page: the machines that ran it and their signatures.

buildenginewhose workmemorycontext provedbest tok/smachines
Q4_K_M by googlegemma4:e4bollamavendor · google3.6 GB128K93.71
IQ4_XS by unslothhf.co/unsloth/gemma-4-E4B-it-GGUF:IQ4_XSollamaquantizer · unsloth6.6 GB128K61.41
Q4_K_M by unslothhf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_K_Mollamaquantizer · unsloth6.9 GB128K56.81
Q4_0 by unslothhf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_0ollamaquantizer · unsloth6.7 GB128K54.71
Q8_0 by unslothhf.co/unsloth/gemma-4-E4B-it-GGUF:Q8_0ollamaquantizer · unsloth9.3 GB128K39.51
BF16 by unslothhf.co/unsloth/gemma-4-E4B-it-GGUF:BF16ollamaquantizer · unsloth12.3 GB128K25.81
Q3_K_M by unslothhf.co/unsloth/gemma-4-E4B-it-GGUF:Q3_K_Mollamaquantizer · unsloth—4K20.41
by ollamagemma4:e4b-it-qatollamaollama———1