Private AI
A RP/storywriting specialist model, full-parameter finetune of Qwen2.5-32B on mixture of synthetic and natural data. It uses Celeste 70B 0.1 data mixture, greatly expanding it to improve versatility, creativity and flavor of the resulting model.
Added Jul 26, 2025
Context Window
16.4K
Max Output
8.2K
Input Price (Auto)
$0.70/1M
Output Price (Auto)
$0.70/1M
Cache Read (Auto)
$0.35/1M
Performance metrics and benchmarks
Sourced from LMArena.
Arena Score
1374.1
Overall Rank
#165 / 383
Votes
32,589
Confidence Interval
1370.0 - 1378.2
Coding
#180 / 378
5,094 votes
1402.8
Math
#174 / 370
3,305 votes
1363.4
Longer Query
#158 / 361
4,509 votes
1384.5
Creative Writing
#150 / 381
4,998 votes
1354.0
Instruction Following
#171 / 383
10,953 votes
1357.0
Hard Prompts
#170 / 383
9,658 votes
1384.9
Auto routing is available for this model. Explicit provider selection is not available.
Loading provider options…
French
#131 / 267
328 votes
1409.1
Japanese
#131 / 245
690 votes
1310.2
Spanish
#145 / 269
248 votes
1377.3
German
#148 / 287
759 votes
1354.3
Korean
#150 / 259
465 votes
1316.0
Polish
#154 / 211
862 votes
1358.2
Industry Writing And Literature And Language
#156 / 382
8,191 votes
1362.5
Non English
#160 / 383
13,913 votes
1361.4
Russian
#160 / 346
2,800 votes
1366.3
Industry Legal And Government
#161 / 356
1,866 votes
1386.5
Chinese
#163 / 355
1,950 votes
1396.9
Exclude Ties
#163 / 383
21,958 votes
1349.6
Industry Entertainment And Sports And Media
#163 / 381
5,985 votes
1339.8
Industry Life And Physical And Social Science
#164 / 381
5,693 votes
1389.7
Industry Medicine And Healthcare
#164 / 353
1,594 votes
1394.9
Multi Turn
#165 / 381
4,868 votes
1372.6
Industry Business And Management And Financial Operations
#167 / 376
3,678 votes
1368.7
English
#173 / 383
18,676 votes
1382.2
Industry Mathematical
#174 / 365
2,941 votes
1367.2
Expert
#175 / 332
1,667 votes
1366.0
Industry Software And It Services
#179 / 383
8,618 votes
1397.8
Hard Prompts English
#181 / 382
6,003 votes
1387.7
Published 2026-08-03 · Matched as qwen2.5-max
LMArena Dataset