SeaWolf-AI commited on
Commit
2fc4834
·
verified ·
1 Parent(s): 1556a30

FreeCPU: ~6-7 tok/s

Browse files
Files changed (1) hide show
  1. index.html +269 -269
index.html CHANGED
@@ -1,269 +1,269 @@
1
- <!doctype html>
2
- <html lang="en">
3
- <head>
4
- <meta charset="utf-8">
5
- <meta name="viewport" content="width=device-width, initial-scale=1">
6
- <title>VKUE — VIDRAFT Kernel Ubiquitous Engine</title>
7
- <meta name="description" content="VKUE Efficiency Leaderboard — serve frontier-class models on minimal hardware.">
8
- </head>
9
- <body>
10
- <style>
11
- :root{
12
- --ground:#F4F7F7; --panel:#FFFFFF; --ink:#0B1418; --muted:#5B6B70;
13
- --line:#D7E0E1; --accent:#0E8C9E; --accent-soft:#E2F4F6; --hero:#B0791A;
14
- --hero-soft:#FBF1DE; --good:#1F8A5B; --good-soft:#E4F3EC; --star:#0E8C9E;
15
- --mono:ui-monospace,"SFMono-Regular","Cascadia Code","Consolas",monospace;
16
- --sans:-apple-system,BlinkMacSystemFont,"Segoe UI","Apple SD Gothic Neo","Malgun Gothic",Roboto,sans-serif;
17
- }
18
- @media (prefers-color-scheme:dark){
19
- :root{
20
- --ground:#0B1418; --panel:#111E23; --ink:#E6EFF0; --muted:#8CA1A6;
21
- --line:#22343A; --accent:#22B8CF; --accent-soft:#0F2A31; --hero:#E0A83E;
22
- --hero-soft:#2A2113; --good:#43D08A; --good-soft:#10281D; --star:#22B8CF;
23
- }
24
- }
25
- :root[data-theme="light"]{
26
- --ground:#F4F7F7; --panel:#FFFFFF; --ink:#0B1418; --muted:#5B6B70;
27
- --line:#D7E0E1; --accent:#0E8C9E; --accent-soft:#E2F4F6; --hero:#B0791A;
28
- --hero-soft:#FBF1DE; --good:#1F8A5B; --good-soft:#E4F3EC; --star:#0E8C9E;
29
- }
30
- :root[data-theme="dark"]{
31
- --ground:#0B1418; --panel:#111E23; --ink:#E6EFF0; --muted:#8CA1A6;
32
- --line:#22343A; --accent:#22B8CF; --accent-soft:#0F2A31; --hero:#E0A83E;
33
- --hero-soft:#2A2113; --good:#43D08A; --good-soft:#10281D; --star:#22B8CF;
34
- }
35
- *{box-sizing:border-box}
36
- body{margin:0;background:var(--ground);color:var(--ink);font-family:var(--sans);
37
- line-height:1.5;-webkit-font-smoothing:antialiased}
38
- .wrap{max-width:1060px;margin:0 auto;padding:clamp(20px,4vw,52px) clamp(16px,4vw,40px)}
39
- .eyebrow{font-family:var(--mono);font-size:12px;letter-spacing:.22em;text-transform:uppercase;
40
- color:var(--accent);margin:0 0 14px}
41
- .wordmark{font-family:var(--mono);font-weight:700;font-size:clamp(38px,8vw,68px);
42
- letter-spacing:-.01em;margin:0;line-height:.98;text-wrap:balance}
43
- .wordmark small{color:var(--muted);font-weight:400;font-size:.30em;letter-spacing:.02em;display:block;margin-top:12px}
44
- .tagline{font-size:clamp(17px,2.4vw,22px);color:var(--ink);margin:18px 0 0;max-width:52ch;text-wrap:balance}
45
- .tagline b{color:var(--accent)}
46
- .sibs{display:flex;flex-wrap:wrap;gap:10px;margin-top:22px}
47
- .chip{font-family:var(--mono);font-size:12.5px;padding:7px 13px;border:1px solid var(--line);
48
- border-radius:999px;color:var(--muted);background:var(--panel)}
49
- .chip b{color:var(--ink)}
50
- .chip.now{border-color:var(--accent);color:var(--accent)}
51
- .chip.link{text-decoration:none;cursor:pointer;transition:border-color .15s,color .15s}
52
- .chip.link:hover{border-color:var(--accent);color:var(--accent)}
53
- .card a.tlink{text-decoration:none}
54
- .card a.tlink:hover{text-decoration:underline}
55
-
56
- .stats{display:grid;grid-template-columns:repeat(auto-fit,minmax(198px,1fr));gap:14px;margin:36px 0 8px}
57
- .stat{background:var(--panel);border:1px solid var(--line);border-radius:14px;padding:20px 18px}
58
- .stat .n{font-family:var(--mono);font-size:clamp(23px,3.8vw,32px);font-weight:700;letter-spacing:-.02em;
59
- color:var(--accent);font-variant-numeric:tabular-nums}
60
- .stat.hero .n{color:var(--hero)}
61
- .stat .l{font-size:13.5px;color:var(--muted);margin-top:7px}
62
- .stat .l b{color:var(--ink);font-weight:600}
63
-
64
- h2{font-size:13px;font-family:var(--mono);letter-spacing:.14em;text-transform:uppercase;
65
- color:var(--muted);margin:44px 0 4px;font-weight:600}
66
- .h2sub{font-size:14px;color:var(--muted);margin:0 0 16px}
67
- .h2sub b{color:var(--ink)}
68
-
69
- .tablecard{background:var(--panel);border:1px solid var(--line);border-radius:16px;overflow:hidden}
70
- .scroll{overflow-x:auto}
71
- table{border-collapse:collapse;width:100%;min-width:800px;font-size:14px}
72
- thead th{font-family:var(--mono);font-size:11px;letter-spacing:.08em;text-transform:uppercase;
73
- color:var(--muted);text-align:left;padding:14px 16px;border-bottom:1px solid var(--line);font-weight:600;white-space:nowrap}
74
- thead th.r,tbody td.r{text-align:right}
75
- tbody td{padding:14px 16px;border-bottom:1px solid var(--line);vertical-align:middle}
76
- tbody tr:last-child td{border-bottom:none}
77
- .mname{font-weight:600;font-size:14.5px;display:flex;align-items:center;gap:8px;flex-wrap:wrap}
78
- .mname .star{color:var(--star)}
79
- .lineage{font-size:11.5px;color:var(--muted);margin-top:3px;font-family:var(--mono)}
80
- .gpqa{font-family:var(--mono);font-weight:700;font-size:16px;font-variant-numeric:tabular-nums}
81
- .method{font-size:10.5px;color:var(--muted);margin-top:2px;letter-spacing:.01em}
82
- .size{font-family:var(--mono);font-size:12.5px;color:var(--muted);font-variant-numeric:tabular-nums;white-space:nowrap}
83
- .hw{font-size:12.5px;color:var(--muted);white-space:nowrap}
84
- .vkue{font-family:var(--mono);font-weight:600;font-size:13px;color:var(--ink);white-space:nowrap}
85
- .pill{display:inline-block;font-family:var(--mono);font-size:11px;padding:3px 9px;border-radius:999px;
86
- background:var(--accent-soft);color:var(--accent);border:1px solid transparent;white-space:nowrap}
87
- .pill.cpu{background:var(--good-soft);color:var(--good)}
88
- tr.hero td{background:var(--hero-soft)}
89
- tr.hero .gpqa,tr.hero .vkue{color:var(--hero)}
90
- tr.hero .pill{background:transparent;border-color:var(--hero);color:var(--hero)}
91
- tr.base td{opacity:.72}
92
- .tag-hero{font-family:var(--mono);font-size:10px;padding:2px 7px;border-radius:5px;background:var(--hero);
93
- color:#0B1418;font-weight:700;letter-spacing:.04em}
94
-
95
- .legend{display:grid;grid-template-columns:repeat(auto-fit,minmax(240px,1fr));gap:10px 24px;
96
- margin-top:20px;font-size:12.5px;color:var(--muted)}
97
- .legend div b{color:var(--ink);font-weight:600}
98
- .legend .k{font-family:var(--mono);color:var(--accent)}
99
-
100
- .cols{display:grid;grid-template-columns:1fr 1fr;gap:14px;margin-top:16px}
101
- .card{background:var(--panel);border:1px solid var(--line);border-radius:14px;padding:22px}
102
- .card .t{font-family:var(--mono);font-weight:700;font-size:15px;margin:0 0 4px}
103
- .card.a .t{color:var(--hero)} .card.b .t{color:var(--accent)}
104
- .card p{margin:8px 0 0;font-size:13.5px;color:var(--muted)}
105
- .card p b{color:var(--ink)}
106
- .note{margin-top:16px;padding:14px 16px;border:1px dashed var(--line);border-radius:12px;
107
- font-size:12.5px;color:var(--muted);background:var(--panel)}
108
- .note b{color:var(--ink)}
109
- footer{margin-top:40px;padding-top:20px;border-top:1px solid var(--line);
110
- font-size:12px;color:var(--muted);font-family:var(--mono);display:flex;justify-content:space-between;flex-wrap:wrap;gap:8px}
111
- @media (max-width:680px){.stats,.cols{grid-template-columns:1fr}}
112
- </style>
113
-
114
- <div class="wrap">
115
- <p class="eyebrow">VIDRAFT · Inference Engine</p>
116
- <h1 class="wordmark">VKUE<small>VIDRAFT Kernel Ubiquitous Engine</small></h1>
117
- <p class="tagline">No GPU? Runs anyway. <b>Everywhere.</b> The ubiquity engine that serves frontier-class models on minimal hardware.</p>
118
- <div class="sibs">
119
- <a class="chip link" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener"><b>VKAE</b> — Speed &#8599;</a>
120
- <span class="chip now"><b>VKUE</b> — Ubiquity · Anywhere</span>
121
- <a class="chip link" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener"><b>&#9654; Live: GPU vs CPU</b> &#8599;</a>
122
- <a class="chip link" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-CPU" target="_blank" rel="noopener"><b>&#9654; Live: CPU-only</b> &#8599;</a>
123
- <a class="chip link" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-FreeCPU" target="_blank" rel="noopener"><b>&#9654; Live: FREE CPU</b> &#8599;</a>
124
- </div>
125
-
126
- <div class="stats">
127
- <div class="stat hero">
128
- <div class="n">18,057 tok/s</div>
129
- <div class="l">Ourbox-35B on a <b>single B200</b> &mdash; VKAE speed ceiling. <b>Measured</b>, aggregate.</div>
130
- </div>
131
- <div class="stat">
132
- <div class="n">20 tok/s</div>
133
- <div class="l"><b>Same model</b> on an <b>8&nbsp;GB laptop</b> &mdash; VKUE ubiquity floor. <b>Measured</b>, 3.7&times; a dense 32B.</div>
134
- </div>
135
- <div class="stat">
136
- <div class="n">FREE CPU</div>
137
- <div class="l"><b>34.7B</b> on HuggingFace&rsquo;s <b>free public CPU space</b> (2&nbsp;vCPU) &mdash; <b>~5.5 tok/s</b>, zero cost. <a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-FreeCPU" target="_blank" rel="noopener" style="color:var(--accent)">Try it &#8599;</a></div>
138
- </div>
139
- <div class="stat">
140
- <div class="n">744B &rarr; 1 GPU</div>
141
- <div class="l">Even <b>GLM-5.2</b>-class giants on a single <b>H100</b> (normally 6&ndash;8).</div>
142
- </div>
143
- <div class="stat">
144
- <div class="n">90.9%</div>
145
- <div class="l">Darwin-398B-JGOS top <b>GPQA Diamond</b> (greedy, no test-time engine).</div>
146
- </div>
147
- </div>
148
-
149
- <h2>VKUE Efficiency Leaderboard</h2>
150
- <p class="h2sub">How little hardware a given quality needs &mdash; <b>27B and up</b>. GPQA values are each model card&rsquo;s posted figure (method noted).</p>
151
-
152
- <div class="tablecard"><div class="scroll">
153
- <table>
154
- <thead>
155
- <tr>
156
- <th>Model</th>
157
- <th class="r">GPQA Diamond</th>
158
- <th class="r">Size (total / active)</th>
159
- <th>Standard HW</th>
160
- <th>VKUE Min HW</th>
161
- </tr>
162
- </thead>
163
- <tbody>
164
- <tr class="hero">
165
- <td><div class="mname">GLM-5.2 <span class="tag-hero">SYMBOLIC</span></div><div class="lineage">Z.ai · 744B MoE · frontier</div></td>
166
- <td class="r"><div class="gpqa">frontier</div><div class="method">official figure &mdash; citation pending</div></td>
167
- <td class="r"><span class="size">744B / 40B</span></td>
168
- <td><span class="hw">H100 &times;6&ndash;8 (INT4)</span></td>
169
- <td><span class="vkue">H100 &times;1</span> <span class="pill">&asymp; 1/6&ndash;1/8</span><div class="method">capacity verified · speed pending</div></td>
170
- </tr>
171
- <tr>
172
- <td><div class="mname">Darwin-398B-JGOS</div><div class="lineage">Darwin V9 · 397B MoE</div></td>
173
- <td class="r"><div class="gpqa">90.9</div><div class="method">greedy · no engine</div></td>
174
- <td class="r"><span class="size">397B / 17B</span></td>
175
- <td><span class="hw">pure GPU &times;3&ndash;4 (INT4)</span></td>
176
- <td><span class="vkue">1 GPU + RAM</span> <span class="pill">available now</span></td>
177
- </tr>
178
- <tr>
179
- <td><div class="mname">Darwin-28B-REASON</div><div class="lineage">Darwin · DELPHI engine</div></td>
180
- <td class="r"><div class="gpqa">89.39</div><div class="method">maj · DELPHI engine</div></td>
181
- <td class="r"><span class="size">26.9B</span></td>
182
- <td><span class="hw">1 GPU</span></td>
183
- <td><span class="vkue">24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
184
- </tr>
185
- <tr>
186
- <td><div class="mname">Darwin-28B-Opus</div><div class="lineage">Darwin · 28B</div></td>
187
- <td class="r"><div class="gpqa">88.89</div><div class="method">posted figure</div></td>
188
- <td class="r"><span class="size">~28B</span></td>
189
- <td><span class="hw">1 GPU</span></td>
190
- <td><span class="vkue">24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
191
- </tr>
192
- <tr>
193
- <td><div class="mname"><span class="star">&#9733;</span> Darwin-36B-Opus</div><div class="lineage">base &larr; Qwen3.6-35B-A3B</div></td>
194
- <td class="r"><div class="gpqa">88.4</div><div class="method">GPQA Diamond</div></td>
195
- <td class="r"><span class="size">36B / 3B</span></td>
196
- <td><span class="hw">1 GPU (FP8)</span></td>
197
- <td><span class="vkue">consumer 24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
198
- </tr>
199
- <tr>
200
- <td><div class="mname">Darwin-27B-Opus</div><div class="lineage">Qwen3.5 · 27.8B</div></td>
201
- <td class="r"><div class="gpqa">86.9</div><div class="method">posted figure</div></td>
202
- <td class="r"><span class="size">27.8B</span></td>
203
- <td><span class="hw">1 GPU</span></td>
204
- <td><span class="vkue">24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
205
- </tr>
206
- <tr class="hero">
207
- <td><div class="mname"><span class="star">&#9733;</span> Ourbox-35B-JGOS <span class="tag-hero">MEASURED</span></div><div class="lineage">Darwin-36B derivative · <a href="https://huggingface.co/FINAL-Bench/Ourbox-35B-JGOS-GGUF" style="color:inherit">GGUF ↗</a></div></td>
208
- <td class="r"><div class="gpqa">86.4 / 70.7</div><div class="method">maj@8 / greedy (measured)</div></td>
209
- <td class="r"><span class="size">35B / 3B</span></td>
210
- <td><span class="hw">1 GPU (FP8)</span></td>
211
- <td><span class="vkue">A10G 126 &middot; CPU 17 tok/s</span> <a class="pill cpu" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="text-decoration:none">live demo &#8599;</a><div class="method">live measured &middot; also 8GB laptop 20 tok/s (3.7&times; dense 32B)</div></td>
212
- </tr>
213
- <tr>
214
- <td><div class="mname">Darwin-31B-Opus</div><div class="lineage">base &larr; Gemma4-31B · Korean #1</div></td>
215
- <td class="r"><div class="gpqa">85.9</div><div class="method">posted figure</div></td>
216
- <td class="r"><span class="size">32.7B</span></td>
217
- <td><span class="hw">1 GPU</span></td>
218
- <td><span class="vkue">consumer 24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
219
- </tr>
220
- <tr class="base">
221
- <td><div class="mname">Qwen3.6-35B-A3B</div><div class="lineage">base model · Darwin-36B parent</div></td>
222
- <td class="r"><div class="gpqa">&mdash;</div><div class="method">official figure &mdash; pending</div></td>
223
- <td class="r"><span class="size">35B / 3B</span></td>
224
- <td><span class="hw">1 GPU (FP8)</span></td>
225
- <td><span class="vkue">consumer 24GB &times;1</span></td>
226
- </tr>
227
- <tr class="base">
228
- <td><div class="mname">Gemma4-31B</div><div class="lineage">base model · Darwin-31B parent</div></td>
229
- <td class="r"><div class="gpqa">&mdash;</div><div class="method">official figure &mdash; pending</div></td>
230
- <td class="r"><span class="size">31B</span></td>
231
- <td><span class="hw">1 GPU</span></td>
232
- <td><span class="vkue">consumer 24GB &times;1</span></td>
233
- </tr>
234
- </tbody>
235
- </table>
236
- </div></div>
237
-
238
- <div class="legend">
239
- <div><span class="k">Method labels</span> &middot; <b>greedy</b> = single pass, <b>maj@8</b> = majority of 8 samples, <b>DELPHI</b> = our test-time engine. Different conditions &mdash; not directly comparable.</div>
240
- <div><span class="k">VKUE Min HW</span> &middot; INT4, held fully in VRAM or served with little / no GPU. <b>CPU-capable</b> = runs without a GPU (slower) thanks to ~3B active params.</div>
241
- <div><span class="k">Throughput (tok/s)</span> &middot; Ourbox-35B measured: <b>18,057</b> aggregate on B200 (<a class="tlink" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener" style="color:var(--accent)">VKAE &#8599;</a>) / <b>126</b> on one A10G &amp; <b>17</b> on a GPU-less CPU box (<a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="color:var(--accent)">live demo &#8599;</a>) / <b>20</b> on an 8GB laptop. All single-stream, generate-format&ndash;gated.</div>
242
- <div><span class="k">Base models</span> &middot; Qwen3.6-35B / Gemma4-31B are our bases; their GPQA figures will be cited from official sources.</div>
243
- </div>
244
-
245
- <h2>Engine Siblings</h2>
246
- <div class="cols">
247
- <div class="card a">
248
- <p class="t"><a class="tlink" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener" style="color:inherit">VKAE &mdash; Speed &#8599;</a></p>
249
- <p><b>How fast.</b> VIDRAFT-optimized serving pushes throughput to the ceiling &mdash; Ourbox-35B <b>measured at 18,057 tok/s</b> aggregate on a single <b>B200</b>. <a class="tlink" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener" style="color:var(--accent)">Open the Speed leaderboard &#8599;</a></p>
250
- </div>
251
- <div class="card b">
252
- <p class="t"><a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="color:inherit">VKUE &mdash; Ubiquity &#8599;</a></p>
253
- <p><b>How low the bar.</b> Same 34.7B model, minimal hardware &mdash; <b>live measured 126 tok/s on one A10G</b> vs <b>17 tok/s on a GPU-less CPU box</b> (identical weights, only the hardware differs), and <b>20 tok/s</b> on an 8&nbsp;GB laptop. If you have a card it&rsquo;s fast; if you don&rsquo;t, it still runs. <a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="color:var(--accent)">Try the live GPU&nbsp;vs&nbsp;CPU demo &#8599;</a></p>
254
- </div>
255
- </div>
256
-
257
- <div class="note">
258
- <b>What&rsquo;s open.</b> Quantized models, benchmark figures, and open-tool run instructions are public &mdash; anyone can reproduce the baseline.
259
- VKUE&rsquo;s engine internals are proprietary; this leaderboard reports only the resulting hardware, never the how.
260
- </div>
261
-
262
- <footer>
263
- <span>FINAL-Bench &middot; VIDRAFT</span>
264
- <span>2026 &middot; VKUE v1 preview</span>
265
- </footer>
266
- </div>
267
-
268
- </body>
269
- </html>
 
1
+ <!doctype html>
2
+ <html lang="en">
3
+ <head>
4
+ <meta charset="utf-8">
5
+ <meta name="viewport" content="width=device-width, initial-scale=1">
6
+ <title>VKUE — VIDRAFT Kernel Ubiquitous Engine</title>
7
+ <meta name="description" content="VKUE Efficiency Leaderboard — serve frontier-class models on minimal hardware.">
8
+ </head>
9
+ <body>
10
+ <style>
11
+ :root{
12
+ --ground:#F4F7F7; --panel:#FFFFFF; --ink:#0B1418; --muted:#5B6B70;
13
+ --line:#D7E0E1; --accent:#0E8C9E; --accent-soft:#E2F4F6; --hero:#B0791A;
14
+ --hero-soft:#FBF1DE; --good:#1F8A5B; --good-soft:#E4F3EC; --star:#0E8C9E;
15
+ --mono:ui-monospace,"SFMono-Regular","Cascadia Code","Consolas",monospace;
16
+ --sans:-apple-system,BlinkMacSystemFont,"Segoe UI","Apple SD Gothic Neo","Malgun Gothic",Roboto,sans-serif;
17
+ }
18
+ @media (prefers-color-scheme:dark){
19
+ :root{
20
+ --ground:#0B1418; --panel:#111E23; --ink:#E6EFF0; --muted:#8CA1A6;
21
+ --line:#22343A; --accent:#22B8CF; --accent-soft:#0F2A31; --hero:#E0A83E;
22
+ --hero-soft:#2A2113; --good:#43D08A; --good-soft:#10281D; --star:#22B8CF;
23
+ }
24
+ }
25
+ :root[data-theme="light"]{
26
+ --ground:#F4F7F7; --panel:#FFFFFF; --ink:#0B1418; --muted:#5B6B70;
27
+ --line:#D7E0E1; --accent:#0E8C9E; --accent-soft:#E2F4F6; --hero:#B0791A;
28
+ --hero-soft:#FBF1DE; --good:#1F8A5B; --good-soft:#E4F3EC; --star:#0E8C9E;
29
+ }
30
+ :root[data-theme="dark"]{
31
+ --ground:#0B1418; --panel:#111E23; --ink:#E6EFF0; --muted:#8CA1A6;
32
+ --line:#22343A; --accent:#22B8CF; --accent-soft:#0F2A31; --hero:#E0A83E;
33
+ --hero-soft:#2A2113; --good:#43D08A; --good-soft:#10281D; --star:#22B8CF;
34
+ }
35
+ *{box-sizing:border-box}
36
+ body{margin:0;background:var(--ground);color:var(--ink);font-family:var(--sans);
37
+ line-height:1.5;-webkit-font-smoothing:antialiased}
38
+ .wrap{max-width:1060px;margin:0 auto;padding:clamp(20px,4vw,52px) clamp(16px,4vw,40px)}
39
+ .eyebrow{font-family:var(--mono);font-size:12px;letter-spacing:.22em;text-transform:uppercase;
40
+ color:var(--accent);margin:0 0 14px}
41
+ .wordmark{font-family:var(--mono);font-weight:700;font-size:clamp(38px,8vw,68px);
42
+ letter-spacing:-.01em;margin:0;line-height:.98;text-wrap:balance}
43
+ .wordmark small{color:var(--muted);font-weight:400;font-size:.30em;letter-spacing:.02em;display:block;margin-top:12px}
44
+ .tagline{font-size:clamp(17px,2.4vw,22px);color:var(--ink);margin:18px 0 0;max-width:52ch;text-wrap:balance}
45
+ .tagline b{color:var(--accent)}
46
+ .sibs{display:flex;flex-wrap:wrap;gap:10px;margin-top:22px}
47
+ .chip{font-family:var(--mono);font-size:12.5px;padding:7px 13px;border:1px solid var(--line);
48
+ border-radius:999px;color:var(--muted);background:var(--panel)}
49
+ .chip b{color:var(--ink)}
50
+ .chip.now{border-color:var(--accent);color:var(--accent)}
51
+ .chip.link{text-decoration:none;cursor:pointer;transition:border-color .15s,color .15s}
52
+ .chip.link:hover{border-color:var(--accent);color:var(--accent)}
53
+ .card a.tlink{text-decoration:none}
54
+ .card a.tlink:hover{text-decoration:underline}
55
+
56
+ .stats{display:grid;grid-template-columns:repeat(auto-fit,minmax(198px,1fr));gap:14px;margin:36px 0 8px}
57
+ .stat{background:var(--panel);border:1px solid var(--line);border-radius:14px;padding:20px 18px}
58
+ .stat .n{font-family:var(--mono);font-size:clamp(23px,3.8vw,32px);font-weight:700;letter-spacing:-.02em;
59
+ color:var(--accent);font-variant-numeric:tabular-nums}
60
+ .stat.hero .n{color:var(--hero)}
61
+ .stat .l{font-size:13.5px;color:var(--muted);margin-top:7px}
62
+ .stat .l b{color:var(--ink);font-weight:600}
63
+
64
+ h2{font-size:13px;font-family:var(--mono);letter-spacing:.14em;text-transform:uppercase;
65
+ color:var(--muted);margin:44px 0 4px;font-weight:600}
66
+ .h2sub{font-size:14px;color:var(--muted);margin:0 0 16px}
67
+ .h2sub b{color:var(--ink)}
68
+
69
+ .tablecard{background:var(--panel);border:1px solid var(--line);border-radius:16px;overflow:hidden}
70
+ .scroll{overflow-x:auto}
71
+ table{border-collapse:collapse;width:100%;min-width:800px;font-size:14px}
72
+ thead th{font-family:var(--mono);font-size:11px;letter-spacing:.08em;text-transform:uppercase;
73
+ color:var(--muted);text-align:left;padding:14px 16px;border-bottom:1px solid var(--line);font-weight:600;white-space:nowrap}
74
+ thead th.r,tbody td.r{text-align:right}
75
+ tbody td{padding:14px 16px;border-bottom:1px solid var(--line);vertical-align:middle}
76
+ tbody tr:last-child td{border-bottom:none}
77
+ .mname{font-weight:600;font-size:14.5px;display:flex;align-items:center;gap:8px;flex-wrap:wrap}
78
+ .mname .star{color:var(--star)}
79
+ .lineage{font-size:11.5px;color:var(--muted);margin-top:3px;font-family:var(--mono)}
80
+ .gpqa{font-family:var(--mono);font-weight:700;font-size:16px;font-variant-numeric:tabular-nums}
81
+ .method{font-size:10.5px;color:var(--muted);margin-top:2px;letter-spacing:.01em}
82
+ .size{font-family:var(--mono);font-size:12.5px;color:var(--muted);font-variant-numeric:tabular-nums;white-space:nowrap}
83
+ .hw{font-size:12.5px;color:var(--muted);white-space:nowrap}
84
+ .vkue{font-family:var(--mono);font-weight:600;font-size:13px;color:var(--ink);white-space:nowrap}
85
+ .pill{display:inline-block;font-family:var(--mono);font-size:11px;padding:3px 9px;border-radius:999px;
86
+ background:var(--accent-soft);color:var(--accent);border:1px solid transparent;white-space:nowrap}
87
+ .pill.cpu{background:var(--good-soft);color:var(--good)}
88
+ tr.hero td{background:var(--hero-soft)}
89
+ tr.hero .gpqa,tr.hero .vkue{color:var(--hero)}
90
+ tr.hero .pill{background:transparent;border-color:var(--hero);color:var(--hero)}
91
+ tr.base td{opacity:.72}
92
+ .tag-hero{font-family:var(--mono);font-size:10px;padding:2px 7px;border-radius:5px;background:var(--hero);
93
+ color:#0B1418;font-weight:700;letter-spacing:.04em}
94
+
95
+ .legend{display:grid;grid-template-columns:repeat(auto-fit,minmax(240px,1fr));gap:10px 24px;
96
+ margin-top:20px;font-size:12.5px;color:var(--muted)}
97
+ .legend div b{color:var(--ink);font-weight:600}
98
+ .legend .k{font-family:var(--mono);color:var(--accent)}
99
+
100
+ .cols{display:grid;grid-template-columns:1fr 1fr;gap:14px;margin-top:16px}
101
+ .card{background:var(--panel);border:1px solid var(--line);border-radius:14px;padding:22px}
102
+ .card .t{font-family:var(--mono);font-weight:700;font-size:15px;margin:0 0 4px}
103
+ .card.a .t{color:var(--hero)} .card.b .t{color:var(--accent)}
104
+ .card p{margin:8px 0 0;font-size:13.5px;color:var(--muted)}
105
+ .card p b{color:var(--ink)}
106
+ .note{margin-top:16px;padding:14px 16px;border:1px dashed var(--line);border-radius:12px;
107
+ font-size:12.5px;color:var(--muted);background:var(--panel)}
108
+ .note b{color:var(--ink)}
109
+ footer{margin-top:40px;padding-top:20px;border-top:1px solid var(--line);
110
+ font-size:12px;color:var(--muted);font-family:var(--mono);display:flex;justify-content:space-between;flex-wrap:wrap;gap:8px}
111
+ @media (max-width:680px){.stats,.cols{grid-template-columns:1fr}}
112
+ </style>
113
+
114
+ <div class="wrap">
115
+ <p class="eyebrow">VIDRAFT · Inference Engine</p>
116
+ <h1 class="wordmark">VKUE<small>VIDRAFT Kernel Ubiquitous Engine</small></h1>
117
+ <p class="tagline">No GPU? Runs anyway. <b>Everywhere.</b> The ubiquity engine that serves frontier-class models on minimal hardware.</p>
118
+ <div class="sibs">
119
+ <a class="chip link" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener"><b>VKAE</b> — Speed &#8599;</a>
120
+ <span class="chip now"><b>VKUE</b> — Ubiquity · Anywhere</span>
121
+ <a class="chip link" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener"><b>&#9654; Live: GPU vs CPU</b> &#8599;</a>
122
+ <a class="chip link" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-CPU" target="_blank" rel="noopener"><b>&#9654; Live: CPU-only</b> &#8599;</a>
123
+ <a class="chip link" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-FreeCPU" target="_blank" rel="noopener"><b>&#9654; Live: FREE CPU</b> &#8599;</a>
124
+ </div>
125
+
126
+ <div class="stats">
127
+ <div class="stat hero">
128
+ <div class="n">18,057 tok/s</div>
129
+ <div class="l">Ourbox-35B on a <b>single B200</b> &mdash; VKAE speed ceiling. <b>Measured</b>, aggregate.</div>
130
+ </div>
131
+ <div class="stat">
132
+ <div class="n">20 tok/s</div>
133
+ <div class="l"><b>Same model</b> on an <b>8&nbsp;GB laptop</b> &mdash; VKUE ubiquity floor. <b>Measured</b>, 3.7&times; a dense 32B.</div>
134
+ </div>
135
+ <div class="stat">
136
+ <div class="n">FREE CPU</div>
137
+ <div class="l"><b>34.7B</b> on HuggingFace&rsquo;s <b>free public CPU space</b> (2&nbsp;vCPU) &mdash; <b>~6-7 tok/s</b>, zero cost. <a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-FreeCPU" target="_blank" rel="noopener" style="color:var(--accent)">Try it &#8599;</a></div>
138
+ </div>
139
+ <div class="stat">
140
+ <div class="n">744B &rarr; 1 GPU</div>
141
+ <div class="l">Even <b>GLM-5.2</b>-class giants on a single <b>H100</b> (normally 6&ndash;8).</div>
142
+ </div>
143
+ <div class="stat">
144
+ <div class="n">90.9%</div>
145
+ <div class="l">Darwin-398B-JGOS top <b>GPQA Diamond</b> (greedy, no test-time engine).</div>
146
+ </div>
147
+ </div>
148
+
149
+ <h2>VKUE Efficiency Leaderboard</h2>
150
+ <p class="h2sub">How little hardware a given quality needs &mdash; <b>27B and up</b>. GPQA values are each model card&rsquo;s posted figure (method noted).</p>
151
+
152
+ <div class="tablecard"><div class="scroll">
153
+ <table>
154
+ <thead>
155
+ <tr>
156
+ <th>Model</th>
157
+ <th class="r">GPQA Diamond</th>
158
+ <th class="r">Size (total / active)</th>
159
+ <th>Standard HW</th>
160
+ <th>VKUE Min HW</th>
161
+ </tr>
162
+ </thead>
163
+ <tbody>
164
+ <tr class="hero">
165
+ <td><div class="mname">GLM-5.2 <span class="tag-hero">SYMBOLIC</span></div><div class="lineage">Z.ai · 744B MoE · frontier</div></td>
166
+ <td class="r"><div class="gpqa">frontier</div><div class="method">official figure &mdash; citation pending</div></td>
167
+ <td class="r"><span class="size">744B / 40B</span></td>
168
+ <td><span class="hw">H100 &times;6&ndash;8 (INT4)</span></td>
169
+ <td><span class="vkue">H100 &times;1</span> <span class="pill">&asymp; 1/6&ndash;1/8</span><div class="method">capacity verified · speed pending</div></td>
170
+ </tr>
171
+ <tr>
172
+ <td><div class="mname">Darwin-398B-JGOS</div><div class="lineage">Darwin V9 · 397B MoE</div></td>
173
+ <td class="r"><div class="gpqa">90.9</div><div class="method">greedy · no engine</div></td>
174
+ <td class="r"><span class="size">397B / 17B</span></td>
175
+ <td><span class="hw">pure GPU &times;3&ndash;4 (INT4)</span></td>
176
+ <td><span class="vkue">1 GPU + RAM</span> <span class="pill">available now</span></td>
177
+ </tr>
178
+ <tr>
179
+ <td><div class="mname">Darwin-28B-REASON</div><div class="lineage">Darwin · DELPHI engine</div></td>
180
+ <td class="r"><div class="gpqa">89.39</div><div class="method">maj · DELPHI engine</div></td>
181
+ <td class="r"><span class="size">26.9B</span></td>
182
+ <td><span class="hw">1 GPU</span></td>
183
+ <td><span class="vkue">24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
184
+ </tr>
185
+ <tr>
186
+ <td><div class="mname">Darwin-28B-Opus</div><div class="lineage">Darwin · 28B</div></td>
187
+ <td class="r"><div class="gpqa">88.89</div><div class="method">posted figure</div></td>
188
+ <td class="r"><span class="size">~28B</span></td>
189
+ <td><span class="hw">1 GPU</span></td>
190
+ <td><span class="vkue">24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
191
+ </tr>
192
+ <tr>
193
+ <td><div class="mname"><span class="star">&#9733;</span> Darwin-36B-Opus</div><div class="lineage">base &larr; Qwen3.6-35B-A3B</div></td>
194
+ <td class="r"><div class="gpqa">88.4</div><div class="method">GPQA Diamond</div></td>
195
+ <td class="r"><span class="size">36B / 3B</span></td>
196
+ <td><span class="hw">1 GPU (FP8)</span></td>
197
+ <td><span class="vkue">consumer 24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
198
+ </tr>
199
+ <tr>
200
+ <td><div class="mname">Darwin-27B-Opus</div><div class="lineage">Qwen3.5 · 27.8B</div></td>
201
+ <td class="r"><div class="gpqa">86.9</div><div class="method">posted figure</div></td>
202
+ <td class="r"><span class="size">27.8B</span></td>
203
+ <td><span class="hw">1 GPU</span></td>
204
+ <td><span class="vkue">24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
205
+ </tr>
206
+ <tr class="hero">
207
+ <td><div class="mname"><span class="star">&#9733;</span> Ourbox-35B-JGOS <span class="tag-hero">MEASURED</span></div><div class="lineage">Darwin-36B derivative · <a href="https://huggingface.co/FINAL-Bench/Ourbox-35B-JGOS-GGUF" style="color:inherit">GGUF ↗</a></div></td>
208
+ <td class="r"><div class="gpqa">86.4 / 70.7</div><div class="method">maj@8 / greedy (measured)</div></td>
209
+ <td class="r"><span class="size">35B / 3B</span></td>
210
+ <td><span class="hw">1 GPU (FP8)</span></td>
211
+ <td><span class="vkue">A10G 126 &middot; CPU 17 tok/s</span> <a class="pill cpu" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="text-decoration:none">live demo &#8599;</a><div class="method">live measured &middot; also 8GB laptop 20 tok/s (3.7&times; dense 32B)</div></td>
212
+ </tr>
213
+ <tr>
214
+ <td><div class="mname">Darwin-31B-Opus</div><div class="lineage">base &larr; Gemma4-31B · Korean #1</div></td>
215
+ <td class="r"><div class="gpqa">85.9</div><div class="method">posted figure</div></td>
216
+ <td class="r"><span class="size">32.7B</span></td>
217
+ <td><span class="hw">1 GPU</span></td>
218
+ <td><span class="vkue">consumer 24GB &times;1</span> <span class="pill cpu">CPU-capable</span></td>
219
+ </tr>
220
+ <tr class="base">
221
+ <td><div class="mname">Qwen3.6-35B-A3B</div><div class="lineage">base model · Darwin-36B parent</div></td>
222
+ <td class="r"><div class="gpqa">&mdash;</div><div class="method">official figure &mdash; pending</div></td>
223
+ <td class="r"><span class="size">35B / 3B</span></td>
224
+ <td><span class="hw">1 GPU (FP8)</span></td>
225
+ <td><span class="vkue">consumer 24GB &times;1</span></td>
226
+ </tr>
227
+ <tr class="base">
228
+ <td><div class="mname">Gemma4-31B</div><div class="lineage">base model · Darwin-31B parent</div></td>
229
+ <td class="r"><div class="gpqa">&mdash;</div><div class="method">official figure &mdash; pending</div></td>
230
+ <td class="r"><span class="size">31B</span></td>
231
+ <td><span class="hw">1 GPU</span></td>
232
+ <td><span class="vkue">consumer 24GB &times;1</span></td>
233
+ </tr>
234
+ </tbody>
235
+ </table>
236
+ </div></div>
237
+
238
+ <div class="legend">
239
+ <div><span class="k">Method labels</span> &middot; <b>greedy</b> = single pass, <b>maj@8</b> = majority of 8 samples, <b>DELPHI</b> = our test-time engine. Different conditions &mdash; not directly comparable.</div>
240
+ <div><span class="k">VKUE Min HW</span> &middot; INT4, held fully in VRAM or served with little / no GPU. <b>CPU-capable</b> = runs without a GPU (slower) thanks to ~3B active params.</div>
241
+ <div><span class="k">Throughput (tok/s)</span> &middot; Ourbox-35B measured: <b>18,057</b> aggregate on B200 (<a class="tlink" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener" style="color:var(--accent)">VKAE &#8599;</a>) / <b>126</b> on one A10G &amp; <b>17</b> on a GPU-less CPU box (<a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="color:var(--accent)">live demo &#8599;</a>) / <b>20</b> on an 8GB laptop. All single-stream, generate-format&ndash;gated.</div>
242
+ <div><span class="k">Base models</span> &middot; Qwen3.6-35B / Gemma4-31B are our bases; their GPQA figures will be cited from official sources.</div>
243
+ </div>
244
+
245
+ <h2>Engine Siblings</h2>
246
+ <div class="cols">
247
+ <div class="card a">
248
+ <p class="t"><a class="tlink" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener" style="color:inherit">VKAE &mdash; Speed &#8599;</a></p>
249
+ <p><b>How fast.</b> VIDRAFT-optimized serving pushes throughput to the ceiling &mdash; Ourbox-35B <b>measured at 18,057 tok/s</b> aggregate on a single <b>B200</b>. <a class="tlink" href="https://huggingface.co/spaces/VIDraft/vkae" target="_blank" rel="noopener" style="color:var(--accent)">Open the Speed leaderboard &#8599;</a></p>
250
+ </div>
251
+ <div class="card b">
252
+ <p class="t"><a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="color:inherit">VKUE &mdash; Ubiquity &#8599;</a></p>
253
+ <p><b>How low the bar.</b> Same 34.7B model, minimal hardware &mdash; <b>live measured 126 tok/s on one A10G</b> vs <b>17 tok/s on a GPU-less CPU box</b> (identical weights, only the hardware differs), and <b>20 tok/s</b> on an 8&nbsp;GB laptop. If you have a card it&rsquo;s fast; if you don&rsquo;t, it still runs. <a class="tlink" href="https://huggingface.co/spaces/FINAL-Bench/Ourbox-35B-VKUE-Demo" target="_blank" rel="noopener" style="color:var(--accent)">Try the live GPU&nbsp;vs&nbsp;CPU demo &#8599;</a></p>
254
+ </div>
255
+ </div>
256
+
257
+ <div class="note">
258
+ <b>What&rsquo;s open.</b> Quantized models, benchmark figures, and open-tool run instructions are public &mdash; anyone can reproduce the baseline.
259
+ VKUE&rsquo;s engine internals are proprietary; this leaderboard reports only the resulting hardware, never the how.
260
+ </div>
261
+
262
+ <footer>
263
+ <span>FINAL-Bench &middot; VIDRAFT</span>
264
+ <span>2026 &middot; VKUE v1 preview</span>
265
+ </footer>
266
+ </div>
267
+
268
+ </body>
269
+ </html>