family member
Gemma 3 logo

Gemma 3 270M

292 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it.

7 builds of Gemma 3 270M, 13 signed runs on real machines: different packagers, precisions and engines, each a different file with its own numbers.

292tok/s, fastest machine
$0.74per M tokens · no API sells it
32Kcontext proved by recall
0.2 GBmemory, smallest build
6providers 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. 6 providers have measured Gemma 3 270M so far.Use it in your own room Ask Evo in your gramx and it runs on a machine that proved it, at the metered $0.74 per million tokens. 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 · 7 builds

trained 32K · proved 32K context, by unsloth Q8_0

291.8 tok/s peak, by unsloth 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

$1USD per million tokenstok/sLinux · L4 23 GB · 12 CPUs · 53 GB RAM · 4-bit by unsloth · 291.82 tok/s · $0.7398/MWindows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM · 4-bit by unsloth · 286.08 tok/s · $0.8964/MWindows · Intel 2 GB · 8 CPUs · 16 GB RAM · 4-bit by unsloth · 51.51 tok/s · $1.3816/MWindows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM · 4-bit by unsloth · 41.45 tok/s · $1.717/MLinux · no GPU · 4 CPUs · 31 GB RAM · 4-bit by unsloth · 37.38 tok/s · $0.819/MLinux · no GPU · 4 CPUs · 31 GB RAM · 2-bit by unsloth · 34.58 tok/s · $0.8836/MLinux · no GPU · 4 CPUs · 31 GB RAM · 2-bit by unsloth · 34.48 tok/s · $0.8863/MLinux · no GPU · 4 CPUs · 31 GB RAM · 3-bit by unsloth · 33.41 tok/s · $0.9147/MLinux · no GPU · 4 CPUs · 31 GB RAM · 4-bit by unsloth · 33 tok/s · $0.9274/MLinux · no GPU · 4 CPUs · 31 GB RAM · 8-bit by unsloth · 30.86 tok/s · $0.993/MWindows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM · 4-bit by unsloth · 27.34 tok/s · $6.412/MLinux · no GPU · 4 CPUs · 31 GB RAM · 16-bit (full precision) by unsloth · 23.52 tok/s · $1.3114/M

Same build, different machines

Q4_K_M by unsloth on 6 machines: the fastest decodes 11× as fast as the slowest. One dot per machine, tokens per second.

  1. NVIDIA L4×1292 tok/s
  2. NVIDIA GeForce RTX 3080 Ti×1286 tok/s
  3. Intel(R) UHD Graphics 730×152 tok/s
  4. Intel(R) Iris(R) Xe Graphics×141 tok/s
  5. Linux, CPU only · 4 CPUs×137 tok/s
  6. AMD Radeon(TM) Vega 8 Mobile Graphics×127 tok/s

What it fits on

The smallest build needs about 0.2 GB; 18 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 RAM6 of 7
  2. NVIDIA GeForce RTX 3050 Laptop GPU4 GB on the cardevery build
  3. NVIDIA GeForce GTX 16504 GB on the cardevery build
  4. AMD Radeon(TM) Vega 8 Mobile Graphics7 GB RAMevery build
  5. Windows, CPU only · 12 CPUs7 GB RAMevery build
  6. Apple silicon8 GB RAMevery build
  7. Apple M18 GB RAMevery build
  8. Mac, CPU only · 8 CPUs8 GB RAMevery build
  9. NVIDIA GeForce RTX 3080 Ti12 GB on the cardevery build
  10. Tesla T415 GB on the cardevery build
  11. Intel(R) UHD Graphics 73016 GB RAMevery build
  12. Intel(R) Iris(R) Xe Graphics16 GB RAMevery build
  13. Apple M416 GB RAMevery build
  14. Apple M116 GB RAMevery build
  15. NVIDIA L422 GB on the cardevery build
  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 32K. 11 walks ended because the model forgot, 0 because the machine could not hold the prompt.

machine · build4K8K16K32Kwhy it stopped
Gram ·acbe75Linux, CPU only · 4 CPUs · Q8_0 by unsloth32K recalled at every size asked
Gram ·acbe75Linux, CPU only · 4 CPUs · F16 by unsloth16K the model lost the needle at 32K
Gram ·dae679NVIDIA L4 · Q4_K_M by unsloth— the model lost the needle at 4K
Gram ·84bf54NVIDIA GeForce RTX 3080 Ti · Q4_K_M by unsloth— the model lost the needle at 4K
Gram ·9e3c79Intel(R) UHD Graphics 730 · Q4_K_M by unsloth— the model lost the needle at 4K
Gram ·2e5c02Intel(R) Iris(R) Xe Graphics · Q4_K_M by unsloth— the model lost the needle at 4K
Gram ·acbe75Linux, CPU only · 4 CPUs · Q4_K_M by unsloth— the model lost the needle at 4K
Gram ·acbe75Linux, CPU only · 4 CPUs · IQ2_M by unsloth— the model lost the needle at 4K
Gram ·acbe75Linux, CPU only · 4 CPUs · IQ2_XXS by unsloth— the model lost the needle at 4K
Gram ·acbe75Linux, CPU only · 4 CPUs · IQ3_XXS by unsloth— the model lost the needle at 4K
Gram ·acbe75Linux, CPU only · 4 CPUs · Q4_0 by unsloth— the model lost the needle at 4K
Gram ·d061c9AMD Radeon(TM) Vega 8 Mobile Graphics · Q4_K_M by unsloth— the model lost the needle at 4K

Carbon per million tokens

Lowest build on each machine · 2 of 6 read off a power sensor, the rest modelled from the parts · fewer grams is better

  1. Gram ·acbe75112 g0.85 J/tok · modelled
  2. Gram ·d061c9295 g1.50 J/tok · modelled
  3. Gram ·dae679424 g3.23 J/tok · sensor
  4. Gram ·9e3c79555 g2.82 J/tok · modelled
  5. Gram ·2e5c021.1 kg5.76 J/tok · modelled
  6. Gram ·84bf542.4 kg12.38 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) · India grid average · 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

proved by at least one build: Structured output · Sustained output · Writes code · Reasons about code

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-3-270m-it-GGUF:Q8_0ollamaquantizer · unsloth0.5 GB32K31.31
F16 by unslothhf.co/unsloth/gemma-3-270m-it-GGUF:F16ollamaquantizer · unsloth0.7 GB16K23.51
Q4_K_M by unslothhf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_Mollamaquantizer · unsloth1.1 GB—291.86
IQ2_M by unslothhf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_Mollamaquantizer · unsloth0.2 GB—34.81
IQ2_XXS by unslothhf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_XXSollamaquantizer · unsloth0.2 GB—34.52
IQ3_XXS by unslothhf.co/unsloth/gemma-3-270m-it-GGUF:IQ3_XXSollamaquantizer · unsloth0.2 GB—33.41
Q4_0 by unslothhf.co/unsloth/gemma-3-270m-it-GGUF:Q4_0ollamaquantizer · unsloth0.3 GB—33.01