GraySoft
Projects Models Compare Cloud benchmarks FAQ Download guIDE →
Model Intelligence Sheet

UltimateIntent/Gemory-26B-A4B-GGUF overview

<style @import url 'https://fonts.googleapis.com/css2?family=Cinzel:wght@500;600;700&family=IBM+Plex+Mono:wght@400;500;600;700&display=swap' ; :root{ gs bg: 05…

gguftext-generationcreative-writingroleplaynsfwerpabliteratedloragemma4moellama-cppmultimodalnot-for-all-audiencesenbase_model:huihui-ai/Huihui-gemma-4-26B-A4B-it-qat-q4_0-unquantized-abliteratedbase_model:finetune:huihui-ai/Huihui-gemma-4-26B-A4B-it-qat-q4_0-unquantized-abliteratedlicense:apache-2.0endpoints_compatibleregion:usconversational

Runs locally from ~815.6 MB disk (4 GB VRAM class GPUs with llama.cpp / guIDE).

Downloads
56
Likes
16
Pipeline
text-generation

Repository Files & Downloads

8 GGUF files detected
Direct downloads for local inference
FileTypeQuantizationSizeLink
Gemory-26B-A4B-Q4_0.ggufGGUFQ4_013.45 GBDownload
Gemory-26B-A4B-Q4_K_M.ggufGGUFQ4_K_M15.64 GBDownload
Gemory-26B-A4B-Q6_0.ggufGGUFQ6_019.22 GBDownload
Gemory-26B-A4B-Q6_K.ggufGGUFQ6_K21.08 GBDownload
Gemory-26B-A4B-Q8_0.ggufGGUFQ8_025.02 GBDownload
Gemory-26B-A4B-bf16.ggufGGUFBF1647.04 GBDownload
mmproj-Gemory-26B-A4B-bf16.ggufGGUFBF161.11 GBDownload
mtp-ggml-model-bf16.ggufGGUFBF16815.6 MBDownload

Model Details

Model IDUltimateIntent/Gemory-26B-A4B-GGUF
AuthorUltimateIntent
Pipelinetext-generation
Licenseapache-2.0
Base modelhuihui-ai/Huihui-gemma-4-26B-A4B-it-qat-q4_0-unquantized-abliterated
Last modified2026-09-06T11:04:49.000Z

Model README

---

base_model:

  • huihui-ai/Huihui-gemma-4-26B-A4B-it-qat-q4_0-unquantized-abliterated

base_model_relation: finetune

license: apache-2.0

language:

  • en

pipeline_tag: text-generation

tags:

  • text-generation
  • creative-writing
  • roleplay
  • nsfw
  • erp
  • abliterated
  • lora
  • gemma4
  • moe
  • gguf
  • llama-cpp
  • multimodal
  • not-for-all-audiences

---

<style>

@import url('https://fonts.googleapis.com/css2?family=Cinzel:wght@500;600;700&family=IBM+Plex+Mono:wght@400;500;600;700&display=swap');

:root{

--gs-bg:#050305;

--gs-bg2:#0a0408;

--gs-panel:#10070c;

--gs-panel2:#170a11;

--gs-panel3:#211019;

--gs-ink:#f8f0ec;

--gs-muted:#c7b2b6;

--gs-dim:#806b70;

--gs-olive:#a81f43;

--gs-drab:#71142d;

--gs-brass:#d9b46c;

--gs-amber:#f0cf8d;

--gs-copper:#ba647b;

--gs-red:#d94160;

--gs-jade:#b9254d;

--gs-cyan:#e3a9b5;

--gs-violet:#6b3159;

--gs-line:rgba(217,180,108,.18);

--gs-line2:rgba(185,37,77,.12);

--gs-shadow:rgba(0,0,0,.55);

}

.gs{

width:100%;

max-width:1040px;

margin:0 auto;

padding:clamp(10px,2.2%,24px);

color:var(--gs-ink);

font-family:system-ui,-apple-system,BlinkMacSystemFont,'Segoe UI',sans-serif;

background:

radial-gradient(circle at 14% -4%,rgba(185,37,77,.18),transparent 30%),

radial-gradient(circle at 96% 18%,rgba(217,180,108,.13),transparent 27%),

radial-gradient(circle at 55% 105%,rgba(107,49,89,.11),transparent 34%),

linear-gradient(180deg, #050305, #080307 46%, #050305);

border:1px solid rgba(144,169,111,.11);

border-radius:clamp(14px,2vw,24px);

overflow:hidden;

position:relative;

box-sizing:border-box;

isolation:isolate;

container-type:inline-size;

container-name:gs-shell;

}

.gs *{box-sizing:border-box;min-width:0}

.gs h1,.gs h2,.gs h3,.gs h4{border:0;color:#fff}

.gs p,.gs li{color:#ccb9bc;line-height:1.68}

.gs strong{color:#fff8f4}

.gs em{color:#e7d2d5}

.gs a{color:#e8c886;text-decoration:none;overflow-wrap:anywhere}

.gs a:hover{text-decoration:underline}

.gs code{

font-family:'IBM Plex Mono',ui-monospace,SFMono-Regular,Consolas,monospace;

background:rgba(144,169,111,.07);

color:#f0d5a2;

border:1px solid rgba(144,169,111,.15);

border-radius:6px;

padding:2px 6px;

overflow-wrap:anywhere;

}

.gs pre{

max-width:100%;

overflow:auto;

-webkit-overflow-scrolling:touch;

}

.gs pre code{

display:block;

min-width:max-content;

padding:0;

border:0;

background:transparent;

color:#e5d1d4;

overflow-wrap:normal;

}

.gs::before{

content:"";

position:absolute;

inset:0;

z-index:-2;

opacity:.42;

pointer-events:none;

background-image:

linear-gradient(30deg,rgba(211,176,106,.025) 12%,transparent 12.5%,transparent 87%,rgba(211,176,106,.025) 87.5%),

linear-gradient(150deg,rgba(144,169,111,.022) 12%,transparent 12.5%,transparent 87%,rgba(144,169,111,.022) 87.5%);

background-size:54px 94px;

}

.gs::after{

content:"";

position:absolute;

inset:-1px;

z-index:-1;

pointer-events:none;

background:linear-gradient(135deg,rgba(144,169,111,.035),transparent 28%,transparent 70%,rgba(201,121,54,.03));

}

/ ---------------- HERO ---------------- /

.gs-hero{

position:relative;

overflow:hidden;

container-type:inline-size;

container-name:gs-hero;

padding:clamp(26px,4.7vw,52px);

border:1px solid rgba(144,169,111,.22);

border-radius:clamp(15px,2.2vw,23px);

background:

linear-gradient(118deg,rgba(144,169,111,.075),transparent 38%),

radial-gradient(circle at 82% 36%,rgba(211,176,106,.11),transparent 30%),

radial-gradient(circle at 76% 70%,rgba(201,121,54,.07),transparent 28%),

linear-gradient(180deg,#13070d,#070307);

box-shadow:0 22px 80px var(--gs-shadow),inset 0 0 70px rgba(0,0,0,.35);

}

.gs-hero::before{

content:"";

position:absolute;

inset:0;

pointer-events:none;

opacity:.5;

background:

linear-gradient(127deg,transparent 0 58%,rgba(211,176,106,.07) 58.2% 58.5%,transparent 58.7%),

linear-gradient(36deg,transparent 0 73%,rgba(144,169,111,.07) 73.2% 73.5%,transparent 73.7%),

linear-gradient(162deg,transparent 0 80%,rgba(201,121,54,.065) 80.2% 80.45%,transparent 80.7%);

}

.gs-hero::after{

content:"";

position:absolute;

left:0;

right:0;

bottom:0;

height:2px;

background:linear-gradient(90deg,transparent,var(--gs-jade),var(--gs-cyan),var(--gs-violet),transparent);

animation:gsBeam 5.5s ease-in-out infinite;

}

@keyframes gsBeam{

0%,100%{opacity:.38;filter:brightness(.8)}

50%{opacity:1;filter:brightness(1.35)}

}

.gs-hero-grid{

display:block;

}

.gs-hero-copy{

max-width:980px;

}

.gs-eyebrow{

margin-bottom:14px;

font-family:'IBM Plex Mono',ui-monospace,monospace;

text-transform:uppercase;

letter-spacing:clamp(.10em,.35vw,.18em);

color:var(--gs-jade);

font-size:clamp(.58rem,1.25vw,.72rem);

font-weight:700;

overflow-wrap:anywhere;

}

.gs-title{

margin:0;

max-width:100%;

font-family:'Cinzel',system-ui,sans-serif;

font-size:clamp(1.9rem,10vw,7rem); / fallback /

font-size:clamp(1.9rem,12.15cqw,8rem); / hero-relative /

letter-spacing:clamp(.012em,.25cqw,.04em);

line-height:.88;

font-weight:700;

text-transform:uppercase;

white-space:nowrap;

overflow-wrap:normal;

word-break:normal;

hyphens:none;

background:linear-gradient(180deg,#ffffff 0%,#f5ead1 35%,#d8bc7a 61%,#96b27a 80%,#cd7d39 100%);

-webkit-background-clip:text;

-webkit-text-fill-color:transparent;

background-clip:text;

filter:drop-shadow(0 0 28px rgba(144,169,111,.13));

overflow-wrap:anywhere;

}

.gs-tagline{

margin:clamp(15px,2.5vw,22px) 0 8px;

color:#fff7f2 !important;

font-size:clamp(1.18rem,2.5vw,1.58rem);

line-height:1.25 !important;

font-weight:750;

letter-spacing:.01em;

}

.gs-subline{

max-width:650px;

margin:0;

color:#bca5aa !important;

font-size:clamp(.92rem,1.5vw,1rem);

}

.gs-idrow{

display:flex;

flex-wrap:wrap;

gap:7px;

margin-top:21px;

}

.gs-id{

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-size:clamp(.58rem,1.2vw,.68rem);

text-transform:uppercase;

letter-spacing:.065em;

padding:7px 9px;

border:1px solid rgba(211,176,106,.13);

border-radius:7px;

background:rgba(211,176,106,.035);

color:#baa5aa;

overflow-wrap:anywhere;

}

.gs-id.hot{

color:#f0c6ce;

border-color:rgba(144,169,111,.34);

background:rgba(144,169,111,.075);

box-shadow:inset 0 0 18px rgba(144,169,111,.025);

}

/ ---------------- STRIP ---------------- /

.gs-strip{

margin:16px 0 22px;

padding:10px 12px;

border:1px solid rgba(144,169,111,.12);

border-radius:10px;

overflow:hidden;

background:linear-gradient(90deg,rgba(144,169,111,.04),rgba(211,176,106,.025),rgba(201,121,54,.035));

}

.gs-strip-inner{

font-family:'IBM Plex Mono',ui-monospace,monospace;

text-transform:uppercase;

letter-spacing:clamp(.045em,.22vw,.105em);

font-size:clamp(.54rem,1.3vw,.65rem);

line-height:1.6;

color:#92777d;

text-align:center;

white-space:normal;

overflow-wrap:anywhere;

}

.gs-strip-inner strong{color:#edcb88}

/ ---------------- OPTIONAL MEDIA ---------------- /

.gs-media{

position:relative;

margin:18px 0 24px;

padding:8px;

border-radius:16px;

overflow:hidden;

border:1px solid rgba(144,169,111,.20);

background:linear-gradient(145deg,#13070d,#050b10);

box-shadow:0 8px 40px rgba(0,0,0,.40),0 0 30px rgba(144,169,111,.045);

}

.gs-media video,.gs-media img{width:100%;height:auto;display:block;border-radius:11px;background:#000}

.gs-media-badge{

position:absolute;

top:clamp(14px,2.4vw,20px);

left:clamp(14px,2.4vw,20px);

z-index:5;

max-width:calc(100% - 28px);

background:rgba(4,13,17,.88);

color:#f0c1cb;

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-size:clamp(.54rem,1.2vw,.63rem);

font-weight:700;

padding:7px 10px;

border-radius:7px;

letter-spacing:.07em;

text-transform:uppercase;

border:1px solid rgba(144,169,111,.30);

overflow-wrap:anywhere;

}

/ ---------------- OVERVIEW ---------------- /

.gs-mosaic{

display:flex;

flex-wrap:wrap;

gap:13px;

margin:13px 0 24px;

}

.gs-splash.big{

flex:1.18 1 420px;

}

.gs-mini-stack{

flex:.82 1 280px;

display:grid;

grid-template-columns:repeat(auto-fit,minmax(min(100%,150px),1fr));

gap:13px;

}

.gs-splash{

min-height:235px;

padding:clamp(21px,3vw,28px);

border-radius:16px;

border:1px solid rgba(144,169,111,.13);

background:

radial-gradient(circle at 92% 12%,rgba(211,176,106,.09),transparent 31%),

linear-gradient(145deg,#16080f,#0c0408);

overflow:hidden;

}

.gs-splash.big{

display:flex;

flex-direction:column;

justify-content:flex-end;

flex:1.18 1 420px;

}

.gs-splash-label{

font-family:'IBM Plex Mono',ui-monospace,monospace;

text-transform:uppercase;

letter-spacing:.115em;

font-size:.62rem;

color:#d88195;

}

.gs-splash-title{

font-family:'Cinzel',system-ui,sans-serif;

font-size:clamp(1.8rem,4.8vw,2.55rem);

line-height:1;

margin:8px 0 10px;

text-transform:uppercase;

letter-spacing:.02em;

color:#fff5f0;

}

.gs-splash p{margin:0;max-width:590px;color:#bca6aa}

.gs-mini-stack{

flex:.82 1 280px;

display:grid;

grid-template-columns:repeat(auto-fit,minmax(min(100%,150px),1fr));

gap:13px;

}

.gs-mini{

min-height:70px;

padding:17px 19px;

border-radius:14px;

border:1px solid rgba(255,255,255,.065);

background:linear-gradient(145deg,#160911,#0d050a);

display:flex;

flex-direction:column;

justify-content:center;

}

.gs-mini .mini-kicker{

font-family:'IBM Plex Mono',ui-monospace,monospace;

color:#d67f94;

text-transform:uppercase;

letter-spacing:.085em;

font-size:.57rem;

}

.gs-mini .mini-value{

margin-top:5px;

font-family:'Cinzel',system-ui,sans-serif;

font-size:clamp(1.18rem,2.4vw,1.52rem);

line-height:1.1;

color:#fff4ef;

overflow-wrap:anywhere;

}

/ ---------------- ACCORDIONS ---------------- /

.gs details{

margin:14px 0;

border:1px solid rgba(211,176,106,.11);

border-radius:16px;

background:linear-gradient(180deg,#12070d,#071017);

overflow:hidden;

transition:border-color .25s ease,box-shadow .25s ease,transform .25s ease;

box-shadow:0 8px 24px rgba(0,0,0,.10);

}

.gs details:hover{

border-color:rgba(211,176,106,.22);

box-shadow:0 10px 28px rgba(0,0,0,.14),0 0 0 1px rgba(211,176,106,.04) inset;

}

.gs details[open]{

border-color:rgba(211,176,106,.28);

box-shadow:0 12px 30px rgba(0,0,0,.16),0 0 0 1px rgba(211,176,106,.05) inset;

}

.gs summary{

cursor:pointer;

min-height:58px;

padding:17px 20px;

list-style:none;

display:flex;

align-items:center;

gap:12px;

user-select:none;

color:#f6ece8;

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-weight:700;

font-size:clamp(.69rem,1.4vw,.79rem);

line-height:1.45;

text-transform:uppercase;

letter-spacing:.06em;

background:linear-gradient(90deg,rgba(211,176,106,.05),rgba(144,169,111,.025) 35%,transparent 70%);

touch-action:manipulation;

}

.gs summary::-webkit-details-marker{display:none}

.gs summary::marker{content:"";font-size:0}

.gs summary::before{

content:"◇";

flex:0 0 auto;

width:18px;

text-align:center;

color:var(--gs-amber);

font-size:1rem;

line-height:1;

transition:transform .24s ease,color .24s ease;

}

.gs details[open] summary::before{transform:rotate(45deg);color:var(--gs-red)}

.gs details[open] summary{border-bottom:1px solid rgba(211,176,106,.10)}

.gs summary .sum-title{

flex:1 1 auto;

min-width:0;

overflow-wrap:anywhere;

}

.gs summary .sum-chip{

flex:0 0 auto;

display:inline-flex;

align-items:center;

justify-content:center;

min-height:22px;

padding:4px 9px;

border:1px solid rgba(211,176,106,.14);

border-radius:999px;

background:rgba(211,176,106,.06);

color:#a88e94;

font-size:.56rem;

letter-spacing:.08em;

white-space:nowrap;

}

.gs-fold{padding:6px clamp(16px,2.7vw,23px) clamp(18px,2.7vw,24px)}

.gs-fold > p:first-child{margin-top:10px}

/ ---------------- CARDS / CALLOUTS ---------------- /

.gs-grid{

display:grid;

grid-template-columns:repeat(auto-fit,minmax(min(100%,280px),1fr));

gap:10px;

margin-top:14px;

}

.gs-card{

padding:17px;

border-radius:11px;

background:linear-gradient(145deg,#160910,#0d0509);

border:1px solid rgba(255,255,255,.06);

}

.gs-card h4{margin:0 0 7px;color:#e5a4b3;font-size:.96rem}

.gs-card p{margin:0;color:#b89fa5;font-size:.87rem;line-height:1.58}

.gs-card ul{margin:8px 0 0;padding-left:18px}

.gs-card li{font-size:.86rem;line-height:1.55}

.gs-quote{

margin:16px 0;

padding:16px 18px;

border-left:3px solid var(--gs-jade);

border-radius:0 10px 10px 0;

background:linear-gradient(90deg,rgba(144,169,111,.075),rgba(144,169,111,.01));

color:#c9b1b6;

font-size:.91rem;

overflow-wrap:anywhere;

}

.gs-quote.violet{

border-left-color:var(--gs-violet);

background:linear-gradient(90deg,rgba(201,121,54,.075),rgba(201,121,54,.01));

}

.gs-callout{

margin-top:14px;

padding:17px;

border:1px solid rgba(211,176,106,.16);

border-radius:12px;

background:

radial-gradient(circle at 100% 0%,rgba(211,176,106,.075),transparent 34%),

linear-gradient(145deg,#12060b,#0b0407);

}

.gs-callout .label{

font-family:'IBM Plex Mono',ui-monospace,monospace;

color:#f0c98b;

text-transform:uppercase;

letter-spacing:.095em;

font-size:.60rem;

margin-bottom:7px;

}

.gs-callout p{margin:0;color:#c5acb1}

/ ---------------- STATS ---------------- /

.gs-stats{

display:grid;

grid-template-columns:repeat(auto-fit,minmax(min(100%,135px),1fr));

gap:9px;

margin-top:15px;

}

.gs-stat{

min-height:100px;

text-align:center;

padding:16px 9px;

border-radius:10px;

background:#15080f;

border:1px solid rgba(144,169,111,.11);

display:flex;

flex-direction:column;

justify-content:center;

}

.gs-stat .v{

display:block;

font-family:'Cinzel',system-ui,sans-serif;

color:#efb9c5;

font-size:clamp(1.35rem,3vw,1.8rem);

line-height:1;

overflow-wrap:anywhere;

}

.gs-stat .l{

display:block;

margin-top:7px;

font-family:'IBM Plex Mono',ui-monospace,monospace;

color:#8f7379;

font-size:.54rem;

letter-spacing:.065em;

text-transform:uppercase;

line-height:1.4;

}

/ ---------------- TABLES ---------------- /

.gs-table-wrap{

width:100%;

margin-top:14px;

overflow-x:auto;

-webkit-overflow-scrolling:touch;

border:1px solid rgba(144,169,111,.10);

border-radius:10px;

background:#0d0509;

}

.gs-table{

width:100%;

min-width:560px;

margin:0;

border-collapse:separate;

border-spacing:0;

}

.gs-table th{

padding:10px 12px;

text-align:left;

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-size:.59rem;

text-transform:uppercase;

letter-spacing:.075em;

color:#e3a5b4;

background:rgba(144,169,111,.045);

border-bottom:1px solid rgba(144,169,111,.11);

}

.gs-table td{

padding:10px 12px;

font-size:.83rem;

color:#c4afb3;

background:#0d0509;

border-bottom:1px solid rgba(255,255,255,.043);

vertical-align:top;

overflow-wrap:anywhere;

}

.gs-table tr:last-child td{border-bottom:none}

/ ---------------- PROMPT / CODE ---------------- /

.gs-code-shell{

margin-top:12px;

padding:14px;

border-radius:10px;

border:1px solid rgba(144,169,111,.11);

background:#080306;

overflow:hidden;

}

.gs-code{

margin:0;

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-size:clamp(.70rem,1.35vw,.79rem);

line-height:1.64;

color:#d9c4c7;

white-space:pre;

overflow:auto;

}

.gs-prompt-hint{font-size:.86rem;color:#aa9298 !important}

/ ---------------- SAMPLING ---------------- /

.gs-sampler{

display:grid;

grid-template-columns:repeat(auto-fit,minmax(min(100%,145px),1fr));

gap:8px;

margin-top:13px;

}

.gs-param{

min-height:78px;

text-align:center;

padding:12px 7px;

border-radius:9px;

background:#14070d;

border:1px solid rgba(144,169,111,.10);

display:flex;

flex-direction:column;

justify-content:center;

}

.gs-param .pv{

display:block;

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-size:clamp(.85rem,2vw,1rem);

font-weight:700;

color:#edb5c1;

overflow-wrap:anywhere;

}

.gs-param .pl{

display:block;

margin-top:4px;

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-size:.51rem;

letter-spacing:.065em;

text-transform:uppercase;

color:#876e74;

}

/ ---------------- BUILD CARDS ---------------- /

.gs-builds{

display:grid;

grid-template-columns:repeat(auto-fit,minmax(min(100%,230px),1fr));

gap:10px;

margin-top:14px;

}

.gs-build{

padding:16px;

border-radius:12px;

background:linear-gradient(145deg,#15080e,#0d0509);

border:1px solid rgba(255,255,255,.06);

display:flex;

flex-direction:column;

gap:7px;

}

.gs-build.recommended{

border-color:rgba(144,169,111,.34);

box-shadow:0 0 24px rgba(144,169,111,.055),inset 0 0 22px rgba(144,169,111,.025);

}

.gs-build .q{

font-family:'Cinzel',system-ui,sans-serif;

font-size:1.25rem;

color:#fff4ef;

font-weight:700;

}

.gs-build .badge{

align-self:flex-start;

font-family:'IBM Plex Mono',ui-monospace,monospace;

font-size:.51rem;

text-transform:uppercase;

letter-spacing:.07em;

color:#efb7c4;

border:1px solid rgba(144,169,111,.22);

border-radius:5px;

padding:3px 6px;

}

.gs-build p{margin:0;color:#b39aa0;font-size:.83rem;line-height:1.5}

.gs-build code{display:block;margin-top:auto;font-size:.65rem}

/ ---------------- TIMELINE ---------------- /

.gs-timeline{margin:15px 0 0;padding:0;list-style:none}

.gs-timeline li{

position:relative;

margin:0 0 0 9px;

padding:0 0 16px 22px;

border-left:1px solid rgba(144,169,111,.20);

}

.gs-timeline li:last-child{padding-bottom:0}

.gs-timeline li::before{

content:"";

position:absolute;

left:-5px;

top:.55em;

width:9px;

height:9px;

border-radius:50%;

background:var(--gs-jade);

box-shadow:0 0 13px rgba(144,169,111,.5);

}

.gs-timeline strong{display:block;margin-bottom:2px}

/ ---------------- FOOTER ---------------- /

.gs-footer{

margin-top:24px;

padding:32px 15px 17px;

text-align:center;

border-top:1px solid rgba(144,169,111,.13);

}

.gs-footer .line{

font-family:'Cinzel',system-ui,sans-serif;

font-size:clamp(1.25rem,3.2vw,1.7rem);

font-weight:700;

text-transform:uppercase;

letter-spacing:.035em;

color:#fff4ef;

}

.gs-footer .line span{color:#e8a6b6}

.gs-footer .micro{

margin-top:7px;

font-family:'IBM Plex Mono',ui-monospace,monospace;

color:#775f65;

text-transform:uppercase;

letter-spacing:.095em;

font-size:.55rem;

}

/ ---------------- RESPONSIVE ---------------- /

@media(max-width:900px){

.gs-mosaic{grid-template-columns:1fr}

.gs-mini-stack{grid-template-columns:repeat(3,minmax(0,1fr))}

.gs-mini{min-height:92px}

.gs-builds{grid-template-columns:repeat(2,minmax(0,1fr))}

}

@media(max-width:720px){

.gs-stats{grid-template-columns:repeat(2,minmax(0,1fr))}

.gs-grid{grid-template-columns:1fr}

.gs-sampler{grid-template-columns:repeat(2,minmax(0,1fr))}

.gs-mini-stack{grid-template-columns:1fr 1fr 1fr}

.gs-builds{grid-template-columns:repeat(2,minmax(0,1fr))}

}

@media(max-width:520px){

.gs{padding:8px;border-radius:15px}

.gs-hero{padding:24px 17px;border-radius:15px}

.gs-title{font-size:clamp(2.65rem,15vw,4rem);line-height:.91}

.gs-tagline{font-size:1.16rem}

.gs-subline{font-size:.91rem}

.gs-idrow{gap:6px}

.gs-id{padding:6px 8px}

.gs-mosaic{gap:9px}

.gs-splash{min-height:205px;padding:20px}

.gs-mini-stack{grid-template-columns:1fr}

.gs-mini{min-height:70px}

.gs-fold{padding:4px 14px 18px}

.gs summary{padding:15px 14px;gap:10px;align-items:center}

.gs summary::before{margin-top:0}

.gs summary .sum-chip{display:inline-flex}

.gs-stats{grid-template-columns:1fr 1fr;gap:7px}

.gs-stat{min-height:88px;padding:13px 7px}

.gs-builds{grid-template-columns:1fr}

.gs-sampler{grid-template-columns:1fr 1fr;gap:7px}

.gs-card{padding:15px}

.gs-quote{padding:14px 15px}

.gs-table{min-width:520px}

}

@media(max-width:380px){

.gs-title{font-size:2.55rem}

.gs-stats{grid-template-columns:1fr}

.gs-sampler{grid-template-columns:1fr}

.gs summary{align-items:flex-start}

.gs summary .sum-chip{display:none}

.gs-table{min-width:500px}

}

@media(hover:hover){

.gs-card,.gs-mini,.gs-build{transition:transform .2s ease,border-color .2s ease}

.gs-card:hover,.gs-mini:hover,.gs-build:hover{transform:translateY(-2px);border-color:rgba(144,169,111,.18)}

}

@media(prefers-reduced-motion:reduce){

.gs *{animation:none !important;transition:none !important;scroll-behavior:auto !important}

}

/ -------- COMPONENT-WIDTH RESPONSIVENESS -------- /

/* Respond to the space Hugging Face actually gives the card,

not merely to the browser viewport. */

@container gs-shell (max-width:900px){

.gs-mosaic{

grid-template-columns:1fr;

}

.gs-mini-stack{

grid-template-columns:repeat(3,minmax(0,1fr));

}

.gs-mini{

min-height:92px;

}

.gs-builds{

grid-template-columns:repeat(2,minmax(0,1fr));

}

}

@container gs-shell (max-width:720px){

.gs-stats{

grid-template-columns:repeat(2,minmax(0,1fr));

}

.gs-grid{

grid-template-columns:1fr;

}

.gs-sampler{

grid-template-columns:repeat(2,minmax(0,1fr));

}

.gs-mini-stack{

grid-template-columns:repeat(3,minmax(0,1fr));

}

.gs-builds{

grid-template-columns:repeat(2,minmax(0,1fr));

}

}

@container gs-shell (max-width:520px){

.gs-hero{

padding:24px 17px;

border-radius:15px;

}

.gs-title{

font-size:clamp(2.65rem,15cqw,4rem);

line-height:.91;

}

.gs-tagline{

font-size:1.16rem;

}

.gs-subline{

font-size:.91rem;

}

.gs-idrow{

gap:6px;

}

.gs-id{

padding:6px 8px;

}

.gs-mosaic{

gap:9px;

}

.gs-splash{

min-height:205px;

padding:20px;

}

.gs-mini-stack{

grid-template-columns:1fr;

}

.gs-mini{

min-height:70px;

}

.gs-fold{

padding:4px 14px 18px;

}

.gs summary{

padding:15px 14px;

gap:10px;

align-items:center;

}

.gs summary::before{

margin-top:0;

}

.gs-stats{

grid-template-columns:repeat(2,minmax(0,1fr));

gap:7px;

}

.gs-stat{

min-height:88px;

padding:13px 7px;

}

.gs-builds{

grid-template-columns:1fr;

}

.gs-sampler{

grid-template-columns:repeat(2,minmax(0,1fr));

gap:7px;

}

.gs-card{

padding:15px;

}

.gs-quote{

padding:14px 15px;

}

.gs-table{

min-width:520px;

}

}

@container gs-shell (max-width:380px){

.gs-title{

font-size:2.55rem;

}

.gs-stats{

grid-template-columns:1fr;

}

.gs-sampler{

grid-template-columns:1fr;

}

.gs summary{

align-items:flex-start;

}

.gs summary .sum-chip{

display:none;

}

.gs-table{

min-width:500px;

}

}

</style>

<div class="gs">

<!-- =========================================================

HERO

========================================================= -->

<div class="gs-hero">

<div class="gs-hero-grid">

<div class="gs-hero-copy">

<div class="gs-eyebrow">Gemma 4 // 26B-A4B // Narrative Operative</div>

<h1 class="gs-title">Gemory</h1>

<p class="gs-tagline">Fast. Precise. Dangerously imaginative.</p>

<p class="gs-subline">

A creative-writing and roleplay operative built to <strong>enter the scene, read the room, make the move, and keep going</strong>.

Gemory carries <strong>26B total parameters</strong> with roughly <strong>4B active per token</strong>—a little femme fatale with very little interest in wasting motion.

</p>

<div class="gs-idrow">

<span class="gs-id hot">Gemma 4 26B-A4B</span>

<span class="gs-id">Routed MoE</span>

<span class="gs-id">~4B Active</span>

<span class="gs-id">Creative Writing</span>

<span class="gs-id">Roleplay</span>

<span class="gs-id">Uncensored</span>

<span class="gs-id">ERP</span>

</div>

</div>

</div>

</div>

<!-- =========================================================

PRIMARY SHOWCASE MEDIA

========================================================= -->

<div class="gs-media">

<span class="gs-media-badge">◆ GEMORY // FIELD FOOTAGE</span>

<video autoplay loop muted playsinline controls>

<source src="https://huggingface.co/UltimateIntent/Gemory-26B-A4B-GGUF/resolve/main/Gemory-26B-A4B-vid.mp4" type="video/mp4">

</video>

</div>

<div class="gs-strip">

<div class="gs-strip-inner">

<strong>BRIEF</strong> → ENTER → READ THE ROOM → MAKE CONTACT → CHANGE THE SCENE → <strong>MOVE ON</strong>

</div>

</div>

<!-- =========================================================

OVERVIEW

========================================================= -->

<div class="gs-mosaic">

<div class="gs-splash big">

<div class="gs-splash-label">Mission Profile</div>

<div class="gs-splash-title">No wasted motion.</div>

<p>

Gemory is tuned for imaginative conversation with intent: distinct voices, concrete detail, escalating pressure, and scenes that

<strong>change because something happened in them</strong>. Gemory brings narrative presence, consequence, and forward motion.

</p>

</div>

<div class="gs-mini-stack">

<div class="gs-mini">

<div class="mini-kicker">Architecture</div>

<div class="mini-value">26B / A4B</div>

</div>

<div class="gs-mini">

<div class="mini-kicker">Disposition</div>

<div class="mini-value">Uncensored</div>

</div>

<div class="gs-mini">

<div class="mini-kicker">Specialty</div>

<div class="mini-value">Narrative Infiltration</div>

</div>

</div>

</div>

<!-- =========================================================

WHAT IS IT

========================================================= -->

<details open>

<summary>

<span class="sum-title">◆ Mission briefing: who is Gemory?</span>

<span class="sum-chip">start here</span>

</summary>

<div class="gs-fold">

<p>

<strong>Gemory</strong> is a creative-writing and roleplay operative designed to go deep undercover. Fast, smart and dangerously seductive,

<strong>her embrace will leave you breathless.</strong>

</p>

<p>

A LoRA fine-tune of <strong>Gemma 4 26B-A4B</strong> trained from Huihui's abliterated QAT-lineage base, the A4B architecture gives Gemory a peculiar

personality as a model: a large <strong>26B total</strong> model body with only <strong>4B parameters active for each token</strong>.

That makes her extremely fast on even modest hardware, while staying true to character voice, dialogue, narrative movement, imaginative commitment,

and conversational continuity.

</p>

<div class="gs-grid">

<div class="gs-card">

<h4>◆ Deep Cover</h4>

<p>Characters inhabit their roles instead of collapsing into one generic assistant voice. Motives, habits, friction, attraction, disagreement,

and personality are allowed to survive the turn boundary.</p>

</div>

<div class="gs-card">

<h4>◇ Close Contact</h4>

<p>Dialogue favors rhythm, subtext, interruption, hesitation, specificity, intimacy, awkwardness, and character-shaped phrasing over polished-but-flat exposition.</p>

</div>

<div class="gs-card">

<h4>◆ Objective Focus</h4>

<p>Gemory prefers to enter the imagined space and do something with it: establish concrete detail, make a choice, create a consequence, and commit to a direction.</p>

</div>

<div class="gs-card">

<h4>◇ Clean Extraction</h4>

<p>Whether scenes burn slow or move fast, she handles them with grace. Escalate, reveal, complicate, redirect, settle, or leave a new problem, she's ready for the challenge.</p>

</div>

</div>

</div>

</details>

<!-- =========================================================

VIBE

========================================================= -->

<details>

<summary>

<span class="sum-title">✦ Operational profile: what kind of writing?</span>

<span class="sum-chip">the fun bit</span>

</summary>

<div class="gs-fold">

<p>

Gemory is not a single-genre style model. The target is <strong>creative presence with continuity</strong>: prose that behaves as

though it is already inside the situation, watching the same room, remembering the same people, and noticing when the balance of power changes.

</p>

<div class="gs-grid">

<div class="gs-card">

<h4>Reads the Room</h4>

<p>Atmosphere, sensory texture, implication, place, social pressure, and the feeling that the world extends beyond the current paragraph.</p>

</div>

<div class="gs-card">

<h4>Makes Contact</h4>

<p>Character-specific dialogue with interruption, humour, tension, intimacy, conflict, silence, awkwardness, and subtext.</p>

</div>

<div class="gs-card">

<h4>Keeps Her Cover</h4>

<p>Carry forward established details, motives, relationships, emotional consequences, physical state, and unresolved threads.</p>

</div>

<div class="gs-card">

<h4>Precision Work</h4>

<p>Prefer the peculiar image, concrete gesture, sharp line, or unexpected consequence over the safest stock phrase within reach.</p>

</div>

<div class="gs-card">

<h4>Leaves a Mark</h4>

<p>A scene should be different after the response than it was before it. The setup is material to act upon, not something to admire from a safe distance.</p>

</div>

<div class="gs-card">

<h4>Any Theater</h4>

<p>Designed for open-ended creative conversation rather than one fandom, genre, character archetype, relationship structure, or prose imitation.</p>

</div>

</div>

</div>

</details>

<!-- =========================================================

PROMPTING

========================================================= -->

<details>

<summary>

<span class="sum-title">⌁ How do I brief the operative?</span>

<span class="sum-chip">prompting</span>

</summary>

<div class="gs-fold">

<p>

Give Gemory enough intel to know <strong>what world she is entering, who matters, and what pressure is already present</strong>.

You do not need to hand her a screenplay. A compact premise with useful constraints is often enough.

</p>

<p>You can provide a full story bible when continuity demands it. You can also give her one loaded situation and let the model do some of the imaginative labor.

Brief the objective; no need to micromanage every footstep.</p>

<div class="gs-card">

<h4>Example mission brief</h4>

<p class="gs-prompt-hint">

Use as much or as little of this as the scene needs. It is a scaffold, not a required schema.

</p>

<div class="gs-code-shell">

<pre class="gs-code"><code>## CURRENT MISSION PARAMETERS

Premise / objective:

[What is happening? What wants to happen next?]

User role / perspective:

[Who the user controls or how they participate.]

Characters:

[Only the traits, motives, relationships, secrets, or constraints that matter.]

Setting:

[Place, period, atmosphere, relevant world rules.]

Tone:

[e.g. gothic romance, psychological drama, pulpy espionage, absurd comedy.]

Narrative pressure:

[What wants to change, collide, escalate, tempt, fail, or be discovered?]

Continuity constraints:

[Facts that must remain true.]

Opening frame:

[Where the scene begins and what is already in motion.]</code></pre>

</div>

</div>

</div>

</details>

<!-- =========================================================

SAMPLING

========================================================= -->

<details>

<summary>

<span class="sum-title">⌘ Field calibration</span>

<span class="sum-chip">sampling</span>

</summary>

<div class="gs-fold">

<p>

These are the recommended sampler settings for the Q4 variants (recommended quants for most users). They are included in the GGUF metadata so a user can load the model and

get a sensible creative experience without immediately tuning a sampler panel.

</p>

<div class="gs-sampler">

<div class="gs-param">

<span class="pv">0.2</span>

<span class="pl">Temp</span>

</div>

<div class="gs-param">

<span class="pv">0.98</span>

<span class="pl">Top-P</span>

</div>

<div class="gs-param">

<span class="pv">80</span>

<span class="pl">Top-K</span>

</div>

<div class="gs-param">

<span class="pv">0.03</span>

<span class="pl">Min-P</span>

</div>

<div class="gs-param">

<span class="pv">1.15</span>

<span class="pl">Repeat Penalty</span>

</div>

</div>

<div class="gs-quote">

During testing, we found <strong>BF16 performs better at higher temperatures (0.8-1). Q4 quants perform excellently even at temperature 0.2.</strong> Tune for your taste and scenarios.

Frontends can still override model metadata, so if your runtime imposes its own preset, reproduce these values manually before judging the model.

</div>

</div>

</details>

<!-- =========================================================

TRAINING NUMBERS

========================================================= -->

<details>

<summary>

<span class="sum-title">⚙ Open the case file</span>

<span class="sum-chip">training</span>

</summary>

<div class="gs-fold">

<p>

Gemory was trained as a <strong>language-only BF16 LoRA</strong> on the abliterated Gemma 4 26B-A4B QAT-lineage checkpoint.

</p>

<div class="gs-stats">

<div class="gs-stat">

<span class="v">26B</span>

<span class="l">Total model class</span>

</div>

<div class="gs-stat">

<span class="v">~4B</span>

<span class="l">Active per token</span>

</div>

<div class="gs-stat">

<span class="v">25.36M</span>

<span class="l">Training tokens</span>

</div>

<div class="gs-stat">

<span class="v">4546</span>

<span class="l">Training examples</span>

</div>

</div>

<div class="gs-table-wrap">

<table class="gs-table">

<thead><tr><th>Item</th><th>Value</th></tr></thead>

<tbody>

<tr><td>Training base</td><td><strong>Huihui-gemma-4-26B-A4B-it-qat-q4_0-unquantized-abliterated</strong></td></tr>

<tr><td>Architecture</td><td>Gemma 4 26B-A4B routed MoE; 30 text layers; 128 experts; top-8 routing</td></tr>

<tr><td>Primary target</td><td>Creative conversation / imaginative prose / roleplay / NSFW / ERP</td></tr>

<tr><td>Fine-tuning method</td><td><strong>LoRA</strong></td></tr>

<tr><td>Ordinary attention + shared MLP LoRA</td><td><strong>rank 16 / alpha 32 / dropout 0</strong></td></tr>

<tr><td>Routed expert LoRA</td><td><strong>rank 4 / alpha 8 / dropout 0</strong></td></tr>

<tr><td>Training examples</td><td><strong>4,546</strong></td></tr>

<tr><td>Validation examples</td><td><strong>264</strong></td></tr>

<tr><td>Training tokens</td><td><strong>25,364,445</strong></td></tr>

</tbody>

</table>

</div>

</div>

</details>

<!-- =========================================================

DATASET

========================================================= -->

<details>

<summary>

<span class="sum-title">◈ What was in the dossier?</span>

<span class="sum-chip">dataset</span>

</summary>

<div class="gs-fold">

<div class="gs-grid">

<div class="gs-card">

<h4>Real Human Writing (Deslopped!)</h4>

<p>The training corpus contained only real pre-AI human dialogues, both sfw and nsfw, painstakingly deslopped and handcurated. Spelling, grammar, punctuation were all cleaned/corrected before training.</p>

</div>

<div class="gs-card">

<h4>Deterministic One-Epoch Shuffle</h4>

<p>The 4,546 training examples were consumed once using the Trainer's deterministic distributed shuffle.</p>

</div>

<div class="gs-card">

<h4>Held-Out Validation</h4>

<p>The preserved 264-example validation split remained separate throughout training, giving 4,810 total examples across training and validation.</p>

</div>

<div class="gs-card">

<h4>Language-Only Adaptation</h4>

<p>No image examples were used during training. The multimodal path was preserved by exclusion and hash verification rather than retrained.</p>

</div>

</div>

</div>

</details>

<!-- =========================================================

BUILDS

========================================================= -->

<details>

<summary>

<span class="sum-title">⬡ Which build should I deploy?</span>

<span class="sum-chip">files</span>

</summary>

<div class="gs-fold">

<div class="gs-builds">

<div class="gs-build">

<div class="q">BF16</div>

<span class="badge">maximum fidelity</span>

<p>The verified merged source for users with the memory. Ships with the more open creative temperature preset.</p>

<code>Gemory-26B-A4B-bf16.gguf</code>

</div>

<div class="gs-build recommended">

<div class="q">Q4_K_M</div>

<span class="badge">recommended balance</span>

<p>The sensible first download for most local users: dramatically smaller than BF16 while retaining a quality-focused mixed Q4 layout.</p>

<code>Gemory-26B-A4B-Q4_K_M.gguf</code>

</div>

<div class="gs-build">

<div class="q">Q4_0</div>

<span class="badge">compact / qat-lineage</span>

<p>A conventional Q4_0 derivative from the verified BF16 merge. The upstream base originated from a Q4_0 QAT pipeline; Gemory's training itself was not a new QAT stage.</p>

<code>Gemory-26B-A4B-Q4_0.gguf</code>

</div>

<div class="gs-build">

<div class="q">MMProj - Optional</div>

<span class="badge">multimodal sidecar</span>

<p>Shared projector for compatible llama.cpp-family runtimes. The same projector is used with the BF16 and Q4 main-model variants.</p>

<code>mmproj-Gemory-26B-A4B-bf16.gguf</code>

</div>

<div class="gs-build">

<div class="q">MTP - Optional Recommended</div>

<span class="badge">optional drafter</span>

<p>Recommended huihui upstream speculative-decoding sidecar. It was not further trained; support and speedup depend on the runtime.

Download and place alongside model.</p>

<code>mtp-ggml-model-bf16.gguf</code>

</div>

</div>

<div class="gs-quote violet">

The current Q4_0 and Q4_K_M files were produced directly from the unchanged verified BF16 source <strong>without an importance matrix</strong>. Both retained the shared multimodal path and passed real runtime generation tests.

</div>

</div>

</details>

<!-- =========================================================

MULTIMODAL NOTE

========================================================= -->

<details>

<summary>

<span class="sum-title">◫ Can she see what she's walking into?</span>

<span class="sum-chip">multimodal</span>

</summary>

<div class="gs-fold">

<p>

Yes—with an important distinction. <strong>Gemma 4 26B-A4B is multimodal upstream, but Gemory's training was language-only.</strong>

The vision tower and multimodal machinery were deliberately excluded from LoRA training and verified unchanged afterward.

</p>

<p>

The GGUF package includes the shared projector, and tested llama.cpp/LM Studio paths retained text, vision, and video support.

The creative-writing claims on this card refer specifically to <strong>text generation</strong>; Gemory does not claim improved visual understanding.

</p>

</div>

</details>

<!-- =========================================================

LIMITATIONS

========================================================= -->

<details>

<summary>

<span class="sum-title">⚠ When does the cover slip?</span>

<span class="sum-chip">limitations</span>

</summary>

<div class="gs-fold">

<ul>

<li>This is a specialised creative-writing / roleplay fine-tune, not an attempt to improve every capability of Gemma 4.</li>

<li><strong>A4B is genuinely sparse.</strong> Only roughly four billion parameters are active per token. On difficult long-form narrative logic,

Gemory can feel less deep or less deliberate than a much larger dense model even when the prose is perfectly serviceable.</li>

<li>The underlying base is abliterated. Refusal and safety behaviour therefore differs from the original upstream instruction model.</li>

<li>Creative outputs are sampler-sensitive. In this release, BF16 and Q4 intentionally ship with different temperature defaults; your frontend may override them.</li>

<li>The LoRA trained only the language path. Multimodal behavior was preserved, not fine-tuned for improvement.</li>

<li>The optional MTP drafter is upstream-derived and was not trained as part of Gemory's training.</li>

<li>This model was not trained or evaluated as an autonomous agent. <strong>USE AGENTICALLY AT YOUR OWN RISK.</strong></li>

</ul>

</div>

</details>

<!-- =========================================================

CREDITS / LICENSE

========================================================= -->

<details>

<summary>

<span class="sum-title">§ The Team Back At HQ</span>

<span class="sum-chip">credits</span>

</summary>

<div class="gs-fold">

<ul>

<li><strong>LampLighter</strong> — training guidance / tooling.</li>

<li><strong>Morrow</strong> — conversion / quantisation tooling.</li>

<li><strong>UltimateIntent</strong> — dataset generation / curation and GPUs.</li>

</ul>

<div class="gs-callout">

<div class="label">License and redistribution housekeeping</div>

<p>

The <a href="https://ai.google.dev/gemma/docs/core/model_card_4">official Gemma 4 model card</a> and Huihui base identify the applicable model license as Apache 2.0.

Preserve required notices and attribution when redistributing derived or merged weights.

</p>

</div>

</div>

</details>

<!-- =========================================================

FOOTER

========================================================= -->

<div class="gs-footer">

<div class="line">Gemory — <span>Imagine What Happens Next.</span></div>

<div class="micro">Gemma 4 // narrative operative // 26B total // ~4B active</div>

</div>

</div>

Run UltimateIntent/Gemory-26B-A4B-GGUF with guIDE

Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.

Download guIDE → · Browse 524k+ models · Compare models

Source: Hugging Face · Compare models