family member

FunctionGemma 270M

341 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 FunctionGemma 270M, 9 signed runs on real machines: different packagers, precisions and engines, each a different file with its own numbers.

341tok/s, fastest machine

$0.63per M tokens · no API sells it

0.2 GBmemory, smallest build

3providers 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. 3 providers have measured FunctionGemma 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 $0.63 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

no context proved on this engine yet

340.8 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

Same build, different machines

Q4_K_M by unsloth on 3 machines: the fastest decodes 14× as fast as the slowest. One dot per machine, tokens per second.

NVIDIA L4×1341 tok/s
Linux, CPU only · 4 CPUs×134 tok/s
Intel(R) Iris(R) Xe Graphics×124 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.

Linux, CPU only · 2 CPUs1 GB RAMevery build
NVIDIA GeForce RTX 3050 Laptop GPU4 GB on the cardevery build
NVIDIA GeForce GTX 16504 GB on the cardevery build
AMD Radeon(TM) Vega 8 Mobile Graphics7 GB RAMevery build
Windows, CPU only · 12 CPUs7 GB RAMevery build
Apple silicon8 GB RAMevery build
Apple M18 GB RAMevery build
Mac, CPU only · 8 CPUs8 GB RAMevery build
NVIDIA GeForce RTX 3080 Ti12 GB on the cardevery build
Tesla T415 GB on the cardevery build
Intel(R) UHD Graphics 73016 GB RAMevery build
Intel(R) Iris(R) Xe Graphics16 GB RAMevery build
Apple M416 GB RAMevery build
Apple M116 GB RAMevery build
NVIDIA L422 GB on the cardevery build
2× Tesla T429 GB on the cardevery build
Linux, CPU only · 4 CPUs31 GB RAMevery build
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. 8 walks ended because the model forgot, 0 because the machine could not hold the prompt.

machine · build	4K	8K	16K	why it stopped
Gram ·dae679 NVIDIA L4 · Q4_K_M by unsloth				— the model lost the needle at 4K
Gram ·acbe75 Linux, CPU only · 4 CPUs · Q4_0 by unsloth				— the model lost the needle at 4K
Gram ·acbe75 Linux, CPU only · 4 CPUs · Q4_K_M by unsloth				— the model lost the needle at 4K
Gram ·acbe75 Linux, CPU only · 4 CPUs · Q8_0 by unsloth				— the model lost the needle at 4K
Gram ·acbe75 Linux, CPU only · 4 CPUs · IQ3_XXS by unsloth				— the model lost the needle at 4K
Gram ·acbe75 Linux, CPU only · 4 CPUs · IQ2_M by unsloth				— the model lost the needle at 4K
Gram ·2e5c02 Intel(R) Iris(R) Xe Graphics · Q4_K_M by unsloth				— the model lost the needle at 4K
Gram ·acbe75 Linux, CPU only · 4 CPUs · BF16 by unsloth				— the model lost the needle at 4K

Carbon per million tokens

Lowest build on each machine · 1 of 3 read off a power sensor, the rest modelled from the parts · fewer grams is better

Gram ·acbe75146 g1.11 J/tok · modelled
Gram ·dae679887 g6.75 J/tok · sensor
Gram ·2e5c021.9 kg9.87 J/tok · modelled

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) · India grid average; 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.

build	engine	whose work	memory	context proved	best tok/s	machines
Q4_K_M by unslothhf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M	ollama	quantizer · unsloth	0.4 GB	—	340.8	3
Q4_0 by unslothhf.co/unsloth/functiongemma-270m-it-GGUF:Q4_0	ollama	quantizer · unsloth	0.3 GB	—	37.2	1
IQ2_XXS by unslothhf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_XXS	ollama	quantizer · unsloth	0.2 GB	—	33.1	1
Q8_0 by unslothhf.co/unsloth/functiongemma-270m-it-GGUF:Q8_0	ollama	quantizer · unsloth	0.4 GB	—	32.9	1
IQ3_XXS by unslothhf.co/unsloth/functiongemma-270m-it-GGUF:IQ3_XXS	ollama	quantizer · unsloth	0.2 GB	—	32.6	1
IQ2_M by unslothhf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_M	ollama	quantizer · unsloth	0.2 GB	—	32.5	1
BF16 by unslothhf.co/unsloth/functiongemma-270m-it-GGUF:BF16	ollama	quantizer · unsloth	0.6 GB	—	18.7	1

derivatives — somebody else's change to these weights

Fine-tunes, merges and abliterations of this member, each its own model with its own page. None of their numbers are counted above.