Gemma 4 E2B
173 tokens a second on Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it.
11 builds of Gemma 4 E2B, 36 signed runs on real machines: different packagers, precisions and engines, each a different file with its own numbers. 5.1B parameters, as the vendor states.
trained 128K · proved 128K context, by unsloth IQ4_XS
174.7 tok/s peak, by unsloth IQ4_XS
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
Same build, different machines
Q4_K_M by google on 9 machines: the fastest decodes 42× as fast as the slowest. One dot per machine, tokens per second.
What it fits on
The smallest build needs about 1.4 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.
- NVIDIA GeForce RTX 3050 Laptop GPU4 GB on the card3 of 6
- NVIDIA GeForce GTX 16504 GB on the card3 of 6
- AMD Radeon(TM) Vega 8 Mobile Graphics7 GB RAM5 of 6
- Windows, CPU only · 12 CPUs7 GB RAM5 of 6
- apple8 GB RAM5 of 6
- Apple M18 GB RAM5 of 6
- NVIDIA GeForce RTX 3080 Ti12 GB on the card5 of 6
- Tesla T415 GB on the card5 of 6
- Intel(R) UHD Graphics 73016 GB RAM5 of 6
- Intel(R) Iris(R) Xe Graphics16 GB RAM5 of 6
- Apple M416 GB RAM5 of 6
- 2× Tesla T429 GB on the cardevery build
- 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, 1 because the machine could not hold the prompt.
| machine · build | 4K | 8K | 16K | 32K | 64K | 128K | why it stopped |
|---|---|---|---|---|---|---|---|
| Gram ·a5be46Tesla T4 · Q4_K_M by google | 128K recalled at every size asked | ||||||
| Kunal HindochaNVIDIA GeForce RTX 3080 Ti · Q4_K_M by unsloth | 128K | ||||||
| Gram ·a5be46Tesla T4 · Q4_0 by unsloth | 128K recalled at every size asked | ||||||
| Gram ·a5be46Tesla T4 · IQ4_XS by unsloth | 128K recalled at every size asked | ||||||
| Gram ·a5be46Tesla T4 · Q4_K_M by unsloth | 128K recalled at every size asked | ||||||
| Gram ·a5be46Tesla T4 · Q8_0 by unsloth | 128K recalled at every size asked | ||||||
| Gram ·a5be46Tesla T4 · BF16 by unsloth | 128K recalled at every size asked | ||||||
| Gram ·e61d3b2× Tesla T4 · Q4_K_M by unsloth | 128K recalled at every size asked | ||||||
| Kunal HindochaNVIDIA GeForce RTX 3080 Ti · IQ4_XS by unsloth | 8K | ||||||
| nitin singhApple M4 · Q4_K_M by google | 8K | ||||||
| nitin singhApple M4 · Q3_K_M by unsloth | 8K | ||||||
| Gram ·e78fd2Tesla T4 · Q4_K_M by unsloth | 4K | ||||||
| Amit KhetanIntel(R) UHD Graphics 730 · Q3_K_M by unsloth | 4K | ||||||
| Amit KhetanIntel(R) UHD Graphics 730 · Q3_K_S by unsloth | 4K |
The 14 longest of 29 walks; every one is on its build's page.
Carbon per million tokens
Lowest build on each machine · 2 of 2 read off a power sensor, the rest modelled from the parts · fewer grams is better
- Gram ·a5be46152 g1.16 J/tok · sensor
- Gram ·e61d3b570 g4.34 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) · world grid average (Singapore not in the table); 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
Amit Khetan11 runs
- Gram ·a5be466 runs
- Gram ·e61d3b5 runs
nitin singh4 runs
Kunal Hindocha3 runs
Subrata Mohapatra1 run
- Gram ·0548a81 run
Keshav Agarwal1 run
- Gram ·e78fd21 run
- Gram ·53b2291 run
and 2 more
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.
| build | engine | whose work | memory | context proved | best tok/s | machines |
|---|---|---|---|---|---|---|
| IQ4_XS by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS | ollama | quantizer · unsloth | 3.9 GB | 128K | 174.7 | 3 |
| Q4_K_M by googlegemma4:e2b | ollama | vendor · google | 1.4 GB | 128K | 170.5 | 9 |
| Q4_K_M by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M | ollama | quantizer · unsloth | 4.9 GB | 128K | 164.7 | 9 |
| Q4_0 by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0 | ollama | quantizer · unsloth | 2.9 GB | 128K | 84.7 | 3 |
| Q8_0 by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0 | ollama | quantizer · unsloth | 6.4 GB | 128K | 67.0 | 2 |
| BF16 by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:BF16 | ollama | quantizer · unsloth | 16.7 GB | 128K | 38.3 | 2 |
| IQ4_NL by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_NL | ollama | quantizer · unsloth | — | 8K | 54.9 | 2 |
| Q3_K_M by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_M | ollama | quantizer · unsloth | — | 8K | 38.4 | 3 |
| Q3_K_S by unslothhf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_S | ollama | quantizer · unsloth | — | 4K | 14.4 | 1 |
| by ollamagemma4:e2b-it-qat | ollama | ollama | — | — | 8.9 | 1 |
| by googlegoogle/gemma-4-E2B-it-qat-q4_0-gguf | llamacpp | — | — | 2.0 | 1 |