family member

Qwen 3.8 27B

2.8 tokens a second on Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM. About $94 per million tokens here, against OpenRouter's $1.88: the API is 50× cheaper. Run it here for privacy or to keep work on your own machines, not to save money.

2 builds of Qwen 3.8 27B, 4 signed runs on real machines: different packagers, precisions and engines, each a different file with its own numbers.

2.9tok/s, fastest machine

$94per M tokens · $1.88 on an API

32Kcontext proved by recall

35.5 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 Qwen 3.8 27B 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 $94 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 · 2 builds

trained 256K · proved 32K context, by unsloth Q8_0

2.9 tok/s peak, by unsloth Q8_0

What it fits on

The smallest build needs about 35.5 GB; 1 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 RAMnone
NVIDIA GeForce RTX 3050 Laptop GPU4 GB on the cardnone
NVIDIA GeForce GTX 16504 GB on the cardnone
AMD Radeon(TM) Vega 8 Mobile Graphics7 GB RAMnone
Windows, CPU only · 12 CPUs7 GB RAMnone
Apple silicon8 GB RAMnone
Apple M18 GB RAMnone
Mac, CPU only · 8 CPUs8 GB RAMnone
NVIDIA GeForce RTX 3080 Ti12 GB on the cardnone
Tesla T415 GB on the cardnone
Intel(R) UHD Graphics 73016 GB RAMnone
Intel(R) Iris(R) Xe Graphics16 GB RAMnone
Apple M416 GB RAMnone
Apple M116 GB RAMnone
NVIDIA L422 GB on the cardnone
2× Tesla T429 GB on the cardnone
Linux, CPU only · 4 CPUs31 GB RAMnone
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, 2 because the machine could not hold the prompt.

machine · build	4K	8K	16K	why it stopped
Gram ·e61d3b 2× Tesla T4 · Q4_K_M by unsloth				8K this machine could not serve at 16K
Gram ·a5be46 2× Tesla T4 · Q8_0 by unsloth				4K the model lost the needle at 8K
Gram ·dae679 NVIDIA L4 · Q4_K_M by unsloth				4K this machine could not serve at 8K

Carbon per million tokens

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

Gram ·e61d3b670 g11.33 J/tok · sensor
Gram ·dae6795.5 kg41.56 J/tok · sensor
Gram ·a5be4622 kg202.09 J/tok · sensor

Carbon is location-based: signed energy × the yearly average of the grid where the machine signed its place (EU grid average (applied to NL) · world grid average (Singapore not in the table) · United States 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: Tool calling · 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
Q8_0 by unslothhf.co/unsloth/Qwen3.8-27B-GGUF:Q8_0	ollama	quantizer · unsloth	35.5 GB	32K	2.9	2
Q4_K_M by unslothunsloth/Qwen3.8-27B-GGUF:Q4_K_M	ollama	quantizer · unsloth	—	8K	—	2

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.