Rohan Paul

for context, these are the max output per response for other top frontier models: > GPT-6 Astra: 128K > Opus 5.5: 128K (300K via Batch API beta) > Fable 5.1: 128K > Grok 4.7: no set cap, but limited by its 500K context > DeepSeek V4 Pro: 384K