The hardware roster · one card
RTX 3080 (10 GB)
10 GB·NVIDIA·graphics card·22 figures, from 2 exhibits · named in 9 exhibits
This roster is in beta and still being tested: every figure is proved against the page it came from, and the way they are read and shown is still changing.
One card carries figures whose record is not on this shelf. 17 of the 22 figures below are bench readings of the RTX 3080: published — on a released page, in a released table, provable word for word against that page’s own row — but every one of them cites a bench file this hub does not publish, and the page itself says so. The other 5 figures are prices, read off a dated public listing their page prints in full and tells you to open for yourself; their record is not this hub’s to publish and it is not missing either, and each of those rows says so on its own line. This card prints both kinds and says which is which rather than implying a receipt that does not exist.
- Language 3
- Pictures 1
- Power, temperature, and what it costs 8
- Everything else this shelf measured for it 10
Narrow these 22 figuresevery modelmodel: a 14-billion model · 1 of 22 figuresmodel: gemma4:26b · 1 of 22 figuresmodel: klein-4b-32st · 2 of 22 figuresmodel: klein-4b-8st · 2 of 22 figuresmodel: klein-4b-8st-pixel4 · 2 of 22 figuresmodel: Mistral Small 3.2 · 2 of 22 figuresmodel: not stated · 12 of 22 figuresevery power cappower cap: 250 W · 13 of 22 figurespower cap: stated in words, not as a number · 1 of 22 figurespower cap: not stated · 8 of 22 figuresevery exhibitexhibit: A Short History of Mistral · 3 of 22 figuresexhibit: Two 3080s against one 3090, and the cap decides · 19 of 22 figures
The rail is CSS and nothing else — no script runs on this page. It narrows by model and by power cap independently, and the line above says what each one is set to; the size of the two together is the one number it cannot count. A group or a band with nothing left in it goes with its rows, so what you are left looking at is what matched. A narrowed view is not part of the address, so it cannot be linked — each card’s own page is a URL, and that is the thing to share.
Language — decode, prefill and first token
Grouped by model and by what was measured, and inside a group sorted by power cap ascending — alphabetical and numeric, never by value. The heading of a group carries everything that does not change inside it; each row carries the value and the conditions that do. A condition the page does not state reads not stated, which is a third thing and not an uncapped card. A † means the same reading is printed in a second place on its exhibit — one reading, two printings, not two figures.
| value | what moves inside this group |
|---|---|
| gemma4:26b decode, tokens a second, split with system memorytok/squantisation, runtime, power cap, context, batch or steps, concurrency and measurement window not stated | |
| 34.40 tokens a second | |
a control, not a like-for-like leg — the page says so in the same sentence, and dates it nowherefrom A Short History of Mistral § What we measured on two cards in one boxtyped in from that page by hand, then checked against it word for word · the card is named in this row · read from a-short-history-of-mistral · section id #what-we-measured-on-two-cards-in-one-box | |
| Mistral Small 3.2 the page says: mistral-small3.2:24b decode split with system memory, against CPU alonecap not set by that harness · 2026-09-10 · quantisation, runtime, context, batch or steps and concurrency not stated | |
| 5.13 tokens a second split, against 2.71 on that box's processor alone — 1.89׆ | |
from A Short History of Mistral § What does it do on our own machines?read off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in this row · read from a-short-history-of-mistral · section id #what-does-it-do-on-our-own-machines | |
| Mistral Small 3.2 the page says: mistral-small3.2:24b decode, tokens a second, split with system memorytok/s2026-09-10 · quantisation, runtime, power cap, context, batch or steps and concurrency not stated | |
| 5.13tok/s† | |
the same reading this page prints in its own table next door, where it carries the date and the bench record it came fromfrom A Short History of Mistral § What we measured on two cards in one boxtyped in from that page by hand, then checked against it word for word · the card is named in the same section · read from a-short-history-of-mistral · section id #what-we-measured-on-two-cards-in-one-box | |
Pictures — seconds per render
1 row — same rule as above.
| value | what moves inside this group |
|---|---|
| not stated what three pictures at a time do to the narrow slotComfyUI · 0.21.1 · 250 W · three pictures at a time · 2026-09-17 · quantisation, context and concurrency not stated | |
| from 1.99 times the x16 board's seconds down to 1.27 | |
the page's own words: on two submissions each, which is thin and is not leant onfrom Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imagetyped in from that page by hand, then checked against it word for word · the card is named in the same section · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
Power, temperature, and what it costs
8 rows — same rule as above.
| value | what moves inside this group |
|---|---|
| not stated core clock while drawing, medianMHzComfyUI · 0.21.1 · 250 W · batch 3 · 2026-09-17 · quantisation, context and concurrency not stated | |
| 1,860 MHz median | |
from Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imagetyped in from that page by hand, then checked against it word for word · the card is named in the same section · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| not stated core temperature peak°CComfyUI · 0.21.1 · 250 W · Ten minutes of continuous drawing · 2026-09-17 · quantisation, context and concurrency not stated | |
| 82 °C | |
the page's own caveat: the two cards did different amounts of work and the single card began the run 11 °C hotter, so this is not a cooler comparison and should not be read as onefrom Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| not stated core temperature plateauComfyUI · 0.21.1 · 250 W · Ten minutes of continuous drawing · 2026-09-17 · quantisation, context and concurrency not stated | |
| 77 °C at 5.6 min | |
the page's own caveat: the two cards did different amounts of work and the single card began the run 11 °C hotter, so this is not a cooler comparison and should not be read as onefrom Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| not stated images drawn in ten minutesimagesComfyUI · 0.21.1 · 250 W · Ten minutes of continuous drawing · 2026-09-17 · quantisation, context and concurrency not stated | |
| 158images | |
the page's own caveat: the two cards did different amounts of work and the single card began the run 11 °C hotter, so this is not a cooler comparison and should not be read as onefrom Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| not stated price, one card, the day's askread between 15:00 and 15:30 UTC on 17 September 2026 · quantisation, runtime, power cap, context, batch or steps and concurrency not stated | |
| $375.00 each | |
the record behind this figure is public — the listing its page prints in full, read in the window that page namesfrom Two 3080s against one 3090, and the cap decides § What it costs, priced 17 September 2026read off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in this row · read from two-used-3080s-priced · section id #what-it-costs-priced-17-september-2026 | |
| not stated price, the EVGA board measured, as listed that dayread between 15:00 and 15:30 UTC on 17 September 2026 · quantisation, runtime, power cap, context, batch or steps and concurrency not stated | |
| $450.00 plus $12.00 shipping | |
the record behind this figure is public — the listing its page prints in full, read in the window that page namesthe page prints this as a note beside the model price and says it is not the price of the rigfrom Two 3080s against one 3090, and the cap decides § What it costs, priced 17 September 2026typed in from that page by hand, then checked against it word for word · the card is named in this row · read from two-used-3080s-priced · section id #what-it-costs-priced-17-september-2026 | |
| not stated price, the Gigabyte board measured, as listed that dayread between 15:00 and 15:30 UTC on 17 September 2026 · quantisation, runtime, power cap, context, batch or steps and concurrency not stated | |
| $620.00 | |
the record behind this figure is public — the listing its page prints in full, read in the window that page namesthe cheaper Gigabyte listings that day being out of stock — the page's own qualificationfrom Two 3080s against one 3090, and the cap decides § What it costs, priced 17 September 2026typed in from that page by hand, then checked against it word for word · the card is named in the same section · read from two-used-3080s-priced · section id #what-it-costs-priced-17-september-2026 | |
| not stated price, the two cardsread between 15:00 and 15:30 UTC on 17 September 2026 · quantisation, runtime, power cap, context, batch or steps and concurrency not stated | |
| $750.00 | |
the record behind this figure is public — the listing its page prints in full, read in the window that page namespriced by MODEL and not by board, because that is what the bench measured and what the marketplace sells — the page's own reasonfrom Two 3080s against one 3090, and the cap decides § What it costs, priced 17 September 2026read off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in this row · read from two-used-3080s-priced · section id #what-it-costs-priced-17-september-2026 | |
Everything else this shelf measured for it
10 rows — same rule as above.
| value | what moves inside this group |
|---|---|
| klein-4b-32st the page says: klein-4b-32st seconds a finished image, median, one card in the x16 slotsComfyUI · 0.21.1 · 250 W · batch 3 · 2026-09-17 · quantisation, context and concurrency not stated | |
| 7.278 s | |
from Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| klein-4b-32st the page says: klein-4b-32st seconds a finished image, median, one card in the x4 slotsComfyUI · 0.21.1 · 250 W · batch 3 · 2026-09-17 · quantisation, context and concurrency not stated | |
| 7.821 s | |
from Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| klein-4b-8st the page says: klein-4b-8st seconds a finished image, median, one card in the x16 slotsComfyUI · 0.21.1 · 250 W · batch 3 · 2026-09-17 · quantisation, context and concurrency not stated | |
| 2.074 s | |
from Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| klein-4b-8st the page says: klein-4b-8st seconds a finished image, median, one card in the x4 slotsComfyUI · 0.21.1 · 250 W · batch 3 · 2026-09-17 · quantisation, context and concurrency not stated | |
| 2.639 s | |
from Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| klein-4b-8st-pixel4 the page says: klein-4b-8st-pixel4 seconds a finished image, median, one card in the x16 slotsComfyUI · 0.21.1 · 250 W · batch 3 · 2026-09-17 · quantisation, context and concurrency not stated | |
| 2.067 s | |
from Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| klein-4b-8st-pixel4 the page says: klein-4b-8st-pixel4 seconds a finished image, median, one card in the x4 slotsComfyUI · 0.21.1 · 250 W · batch 3 · 2026-09-17 · quantisation, context and concurrency not stated | |
| 2.643 s | |
from Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| a 14-billion model residency of a 14-billion model on one card%Q4_K_M · ollama · 0.32.13 · 256 tokens out · Measured 2026-09-17 01:17 to 01:22 UTC · power cap, context and concurrency not stated | |
| 92.1 per cent resident | |
one card of the pair, not the pair: the page's own sentence is about what a 10 GB card holdsfrom Two 3080s against one 3090, and the cap decides § Two copies, or one bigger model?typed in from that page by hand, then checked against it word for word · the card is named elsewhere on that page · read from two-used-3080s-priced · section id #two-copies-or-one-bigger-model | |
| not stated completed sales over the twelve months, average and countover the twelve months to 17 September 2026 · quantisation, runtime, power cap, context, batch or steps and concurrency not stated | |
| $358.00 each across 318 sales | |
the record behind this figure is public — the listing its page prints in full, read in the window that page namesthe page calls this a supporting figure rather than a pricefrom Two 3080s against one 3090, and the cap decides § What it costs, priced 17 September 2026typed in from that page by hand, then checked against it word for word · the card is named in the same section · read from two-used-3080s-priced · section id #what-it-costs-priced-17-september-2026 | |
| not stated fan, maximum%ComfyUI · 0.21.1 · 250 W · Ten minutes of continuous drawing · 2026-09-17 · quantisation, context and concurrency not stated | |
| 92 % | |
the page's own caveat: the two cards did different amounts of work and the single card began the run 11 °C hotter, so this is not a cooler comparison and should not be read as onefrom Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
| not stated seconds an image, mediansComfyUI · 0.21.1 · 250 W · Ten minutes of continuous drawing · 2026-09-17 · quantisation, context and concurrency not stated | |
| 3.737s | |
the page's own caveat: the two cards did different amounts of work and the single card began the run 11 °C hotter, so this is not a cooler comparison and should not be read as onefrom Two 3080s against one 3090, and the cap decides § Renders: one 3080 against one 3090, seconds per imageread off that page’s own table by the roster’s reader, then checked against it word for word · the card is named in the table’s own heading · read from two-used-3080s-priced · section id #renders-one-3080-against-one-3090-seconds-per-image | |
Where it appears
Every released exhibit that names this identity, in the shelf’s own order. A count beside an exhibit is how many of its figures are carried above, and it links the control that narrows the table to them — a number with no way back to what it counts is the thing this shelf exists against.
- A laptop, asked the desktop's questions — named, and no figure carried from it — yet
- A Short History of Mistral — 3 figures on this page
- Electricity per million tokens, card by card — named, and no figure carried from it — yet
- GPUs and orchids — named, and no figure carried from it — yet
- RTX 5080 vs RTX 3080 Ti — named, and no figure carried from it — yet
- Six graphics cards, the same questions — named, and no figure carried from it — yet
- The 3080 Rig, Priced — named, and no figure carried from it — yet
- The same eighteen pictures, card by card — named, and no figure carried from it — yet
- Two 3080s against one 3090, and the cap decides — 19 figures on this page
How this one is named here
The canonical name above is the full product name. Everything below is a spelling a published page or kit has actually used for it, kept so a reader who searches an older page’s words lands on this row. The set is closed and historical: no page cut after 2026-09-17 writes a memory class where it can write a card.
- Dell RTX 3080 (10GB) OEM
- EVGA GeForce RTX 3080 FTW3 Ultra, 10 GB
- GeForce RTX 3080 10 GB
- a 3080 split with system memory
- a used consumer card too small to hold it
- on a 3080 split with system memory
- one 3080
- the EVGA FTW3 Ultra
- the stock EVGA FTW3 Ultra
- two GeForce RTX 3080 10 GB
- two used GeForce RTX 3080 10 GB, any make