BENCHMARK/
the brief, redacted

Build a complete three-tier pricing page as one self-contained HTML document. Tests layout, typographic hierarchy, and restraint under a familiar brief.

HTML pagev1A8A380793653504 chars76 words16 models4 sweeps
00CLAUDE FABLE 5ULTRA
43S·4204TOK·$0.213·9.1KB
object
01GPT-5.6 SOLULTRA
33S·3240TOK·$0.098·10.5KB
object
02GEMINI 3.1 PROULTRA
63S·8922TOK·$0.107·10.6KB
object
03CLAUDE OPUS 5HIGH
90S·8838TOK·$0.222·16.7KB
object
04GPT-5.6 TERRAHIGH
64S·3972TOK·$0.060·12.5KB
object
05CLAUDE SONNET 5MEDIUM
25S·3208TOK·$0.033·6.9KB
object
06GPT-5.6 LUNAMEDIUM
17S·3312TOK·$0.020·10.4KB
object
07GEMINI 3.6 FLASHMEDIUM
31S·7913TOK·$0.060·15.8KB
object
08CLAUDE HAIKU 4.5LOW
13S·2703TOK·$0.014·8.9KB
object
09GPT-5.4 NANOLOW
30S·4452TOK·$0.0056·13.5KB
object
10GEMINI 3.5 FLASH-LITELOW
7.1S·2188TOK·$0.0055·7.7KB
object
11QWEN3 235B A22BLOW
119S·1602TOK·$0.0014·6.5KB
object
12GROK 4.5HIGH
30S·2591TOK·$0.016·7.1KB
object
13KIMI K3HIGH
182S·9591TOK·$0.145·13.3KB
object
14GLM 5.2MEDIUM
15DEEPSEEK V4 PROMEDIUM
40S·3793TOK·$0.0034·12.9KB
object

Search

Prompts, models and categories