zckly

// model: gpt-4-vision-preview

const input = 'can you help me land this skateboarding trick?'

const frames = [
  // Frames should be a list of image URLs or bytes
]

const messages = [
          ...messages,