{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreibfqq6m4jdwhrcou5oddp2mn5gmmvijkumb25pvgjtmcp2d3trhby",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mn6mtcdg7mv2"
  },
  "path": "/t/how-do-you-forecast-what-an-ai-feature-will-actually-cost-before-you-build-it/1382064#post_5",
  "publishedAt": "2026-05-31T21:23:53.000Z",
  "site": "https://community.openai.com",
  "textContent": "But: recent build\n\nGoal: build high quality image alt text for a property image in the vacation rental industry.\n\nInputs: property data payload, high res image url (public)\n\nModel: 5.4-nano\n\nApproach:\n\n  1. Resize image to 1024 largest side\n  2. Extract all visible elements (slightly biased by property description)\n  3. Reprioritize image main focus items based on property booking decision triggers (again biased by property description + main selling features, strongly this time)\n  4. Suggest candidates for image alt text (biased by guest search intent and keyword volume lookup table (static))\n  5. Assemble the final alt text.\n\n\n\nWith prompts, inputs, outputs we are talking about 35K tokens total per run to get:\n\n  1. 5 paragraphs of raw image description (non public result, not usable outside the workflow)\n  2. 3 paragraphs of high quality (commercial) image description (useful as input for room description generation, social media posts generation or image selection)\n  3. 5 alt text candidates (useless if not in later fine-tuning)\n  4. 1 alt text deliverable: 5-12 words\n\n\n\nBasically between cached inputs (static prompt+property description always first) and other API costs we are approx 1 cent per run if you run all images for a single property in parallel (average images per property are 30-60).\n\nSo cost to get all image alt texts per property are around 0.5 USD.\n\nSold at about 2 USD/month/property… wait? Is it X4 only? No, it’s a yearly subscription and properties get shooted around once a year. With some rotation you get somewhere close to X20 between all your costs and the revenue.\n\nFor me, the above math is risky… but I have the distribution channels and at least 200 entreprise accounts + all decision making contacts to sell to. So the risk is acceptable.\n\nHope it exposed some logic I use for costs.",
  "title": "How do you forecast what an AI feature will actually cost — before you build it?"
}