family member
Gemma 4 logo

Gemma 4 12B

Ran on 1 kind of machine here, but its speed was not recorded.

5 builds of Gemma 4 12B, 6 signed runs on real machines: different packagers, precisions and engines, each a different file with its own numbers. 12B parameters, as the vendor states.

29tok/s, fastest machine
256Kcontext proved by recall
8.0 GBmemory, smallest build
2providers measured it
Run it on your gram Install a gram with one command. It works through the catalog on your own machine and signs every result. 2 providers have measured Gemma 4 12B so far.Use it in your own room Ask Evo in your gramx and it runs on a machine that proved it. On your own machines it costs you nothing.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 · 5 builds

trained 256K · proved 256K context, by unsloth Q8_0

29.1 tok/s peak, by unsloth Q4_K_M

What it fits on

The smallest build needs about 8.0 GB; 10 of 18 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. Linux, CPU only · 2 CPUs1 GB RAMnone
  2. NVIDIA GeForce RTX 3050 Laptop GPU4 GB on the cardnone
  3. NVIDIA GeForce GTX 16504 GB on the cardnone
  4. AMD Radeon(TM) Vega 8 Mobile Graphics7 GB RAMnone
  5. Windows, CPU only · 12 CPUs7 GB RAMnone
  6. Apple silicon8 GB RAMnone
  7. Apple M18 GB RAMnone
  8. Mac, CPU only · 8 CPUs8 GB RAMnone
  9. NVIDIA GeForce RTX 3080 Ti12 GB on the card1 of 5
  10. Tesla T415 GB on the card1 of 5
  11. Intel(R) UHD Graphics 73016 GB RAM2 of 5
  12. Intel(R) Iris(R) Xe Graphics16 GB RAM3 of 5
  13. Apple M416 GB RAM3 of 5
  14. Apple M116 GB RAM3 of 5
  15. NVIDIA L422 GB on the card4 of 5
  16. 2× Tesla T429 GB on the cardevery build
  17. Linux, CPU only · 4 CPUs31 GB RAMevery build
  18. NVIDIA RTX PRO 6000 Blackwell Server Edition95 GB on the cardevery 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 256K. 1 walk ended because the model forgot, 1 because the machine could not hold the prompt.

machine · build4K8K16K32K64K128K256Kwhy it stopped
Gram ·a5be462× Tesla T4 · Q8_0 by unsloth256K recalled at every size asked
Gram ·a5be462× Tesla T4 · Q4_K_M by unsloth128K the model lost the needle at 256K
Gram ·a5be462× Tesla T4 · Q4_0 by google128K recalled at every size asked
Gram ·a5be462× Tesla T4 · IQ4_XS by unsloth128K recalled at every size asked
Gram ·a5be462× Tesla T4 · BF16 by unsloth128K too slow to answer here

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
Q8_0 by unslothhf.co/unsloth/gemma-4-12b-it-GGUF:Q8_0ollamaquantizer · unsloth17.6 GB256K18.61
Q4_K_M by unslothhf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_Mollamaquantizer · unsloth15.8 GB128K29.12
Q4_0 by googlehf.co/google/gemma-4-12B-it-qat-q4_0-gguf:Q4_0ollamavendor · google15.6 GB128K27.91
IQ4_XS by unslothhf.co/unsloth/gemma-4-12b-it-GGUF:IQ4_XSollamaquantizer · unsloth8.0 GB128K23.21
BF16 by unslothhf.co/unsloth/gemma-4-12b-it-GGUF:BF16ollamaquantizer · unsloth28.9 GB128K10.71