BENCHMARK/
the brief, redacted

Produce a six-icon weather set on a shared grid, in one SVG. Tests visual consistency across repeated elements — the place where most models drift.

SVGv1CEA26CACF025386 chars64 words16 models4 sweeps
00CLAUDE FABLE 5ULTRA
41S·3045TOK·$0.154·1 PATHS·31 NODES·548B·FILL 9%
51S·3570TOK·$0.181·6 PATHS·14 NODES·437B·FILL 11%
40S·3272TOK·$0.166·5 PATHS·32 NODES·540B·FILL 11%
Sample of 3 for column 03
01GPT-5.6 SOLULTRA
77S·2921TOK·$0.088·6 PATHS·17 NODES·652B·FILL 13%
57S·2899TOK·$0.088·6 PATHS·16 NODES·661B·FILL 10%
43S·1664TOK·$0.051·6 PATHS·18 NODES·597B·FILL 11%
Sample of 3 for column 13
02GEMINI 3.1 PROULTRA
126S·15995TOK·$0.192·5 PATHS·16 NODES·424B·FILL 11%
120S·15996TOK·$0.192
119S·15996TOK·$0.192
Sample of 3 for column 23
03CLAUDE OPUS 5HIGH
130S·10602TOK·$0.266·24 PATHS·39 NODES·662B·FILL 17%
65S·5035TOK·$0.127·12 PATHS·28 NODES·552B·FILL 8%
Sample of 2 for column 32
04GPT-5.6 TERRAHIGH
28S·2368TOK·$0.036·6 PATHS·15 NODES·602B·FILL 9%
17S·1414TOK·$0.022·10 PATHS·18 NODES·570B·FILL 16%
15S·1071TOK·$0.016·10 PATHS·17 NODES·621B·FILL 18%
Sample of 3 for column 43
05CLAUDE SONNET 5MEDIUM
78S·7622TOK·$0.077·2 PATHS·29 NODES·593B·FILL 12%
90S·8708TOK·$0.088·6 PATHS·44 NODES·710B·FILL 67%
81S·7880TOK·$0.079·6 PATHS·44 NODES·670B·FILL 67%
Sample of 3 for column 53
06GPT-5.6 LUNAMEDIUM
16S·1262TOK·$0.0077·10 PATHS·19 NODES·516B·FILL 14%
10S·1191TOK·$0.0073·6 PATHS·18 NODES·642B·FILL 13%
38S·2772TOK·$0.017·7 PATHS·12 NODES·634B·FILL 13%
Sample of 3 for column 63
07GEMINI 3.6 FLASHMEDIUM
38S·9189TOK·$0.069·29 PATHS·36 NODES·670B·FILL 10%
40S·10504TOK·$0.079·23 PATHS·30 NODES·557B·FILL 11%
50S·12227TOK·$0.092·10 PATHS·17 NODES·604B·FILL 8%
Sample of 3 for column 73
08CLAUDE HAIKU 4.5LOW
8.7S·1735TOK·$0.0088·6 PATHS·33 NODES·578B·FILL 11%
10S·1892TOK·$0.0096·6 PATHS·43 NODES·707B·FILL 12%
9.7S·1797TOK·$0.0091·5 PATHS·44 NODES·662B·FILL 8%
Sample of 3 for column 83
09GPT-5.4 NANOLOW
13S·2060TOK·$0.0026·41 PATHS·64 NODES·1.1KB·FILL 16%
13S·1462TOK·$0.0019·35 PATHS·44 NODES·919B·FILL 25%·CLIPPED
11S·1560TOK·$0.0020·35 PATHS·49 NODES·863B·FILL 22%·CLIPPED
Sample of 3 for column 93
10GEMINI 3.5 FLASH-LITELOW
5.8S·2210TOK·$0.0056·8 PATHS·44 NODES·936B·FILL 13%
6.5S·2052TOK·$0.0052·5 PATHS·41 NODES·767B·FILL 18%
5.7S·1909TOK·$0.0048·5 PATHS·40 NODES·783B·FILL 12%
Sample of 3 for column 103
11QWEN3 235B A22BLOW
8.8S·1372TOK·$0.0012·8 PATHS·34 NODES·554B·FILL 13%
9.5S·1323TOK·$0.0012·7 PATHS·33 NODES·580B·FILL 9%
17S·1363TOK·$0.0012·9 PATHS·35 NODES·596B·FILL 11%
Sample of 3 for column 113
12GROK 4.5HIGH
15S·1258TOK·$0.0080·5 PATHS·39 NODES·536B·FILL 14%
18S·1520TOK·$0.0094·5 PATHS·39 NODES·580B·FILL 11%
14S·1125TOK·$0.0070·5 PATHS·35 NODES·508B·FILL 12%
Sample of 3 for column 123
13KIMI K3HIGH
780S
586S·14894TOK·$0.224·6 PATHS·22 NODES·667B·FILL 9%
186S·9850TOK·$0.148·6 PATHS·11 NODES·484B·FILL 7%
Sample of 3 for column 133
14GLM 5.2MEDIUM
0.8S
0.5S
0.3S
Sample of 3 for column 143
15DEEPSEEK V4 PROMEDIUM
196S·16000TOK·$0.014
130S·10510TOK·$0.0092·11 PATHS·35 NODES·534B·FILL 15%
148S·11933TOK·$0.010·10 PATHS·17 NODES·575B·FILL 12%
Sample of 3 for column 153

Search

Prompts, models and categories