Which AI platform to use?

1. Google Gemini API (via Google AI Studio)
Is it permanently free?: Yes.
What are the limits?:

• 15 Requests Per Minute (RPM) (for Gemini 1.5 Flash).
• 1,500 Requests Per Day (RPD).
• Requirement: Just a standard Google Account. You never need to enter a credit card.

2. OpenAI API (GPT Models)

• Is it permanently free?: No. OpenAI does not have an ongoing free tier for API developers.
• What are the limits?:
• New developer accounts sometimes get a one-time $5 free trial credit that expires after 3 months.
• Once those credits are used (or if they expire), the API key stops working, and you must add a credit card to pre-fund your account (minimum $5) to continue using it.

3. Anthropic API (Claude Models)

• Is it permanently free?: No. Anthropic does not have a permanent free tier for their developer API.
• What are the limits?:
• New verified developer accounts receive a one-time $5 free trial credit to test the models.
• Once the $5 is exhausted, you must purchase prepaid credits to keep using the API key.

Summary:

• Gemini is by far the most consumer-friendly option because it is 100% free forever for typical personal photo libraries.
• If you choose to use OpenAI or Anthropic, you will need to pre-pay a few dollars onto your developer accounts to keep those API keys active long-term.

I use Gemini, and decided to get a Google Ultra account (for $100/month) because I develop apps. If you buy a Pixel phone, you get a free year of Gemini Pro, which should provide more than enough AI abilities.

How much does AI cost to use these programs (AI Home Inventory)?

1. Google Gemini (Configured via Google AI Studio Key)

Gemini is currently the most cost-effective platform because of its flat image-to-token pricing.

• Gemini Flash (gemini-3.5-flash / gemini-2.5-flash) — Default / Fast
  • Cost per request: ~$0.00007 (virtually free)
  • Cost to inventory 100 items: Less than 1 cent ($0.007)
• Gemini Pro (gemini-2.5-pro) — High Quality Pro
  • Cost per request: ~$0.0015
  • Cost to inventory 100 items: ~15 cents ($0.15)

2. OpenAI ChatGPT (Configured via OpenAI API Key)

OpenAI's mini model is extremely competitive, while the flagship GPT-4o is premium.

• GPT-4o Mini (gpt-4o-mini) — Cost-efficient
  • Cost per request: ~$0.0003
  • Cost to inventory 100 items: ~3 cents ($0.03)
• GPT-4o (gpt-4o) — High Quality Vision
  • Cost per request: ~$0.005
  • Cost to inventory 100 items: ~50 cents ($0.50)

3. Anthropic Claude (Configured via Anthropic Console Key)

Claude models are excellent at detailed reasoning but have slightly higher pricing because their image token weight is larger.

• Claude 3.5 Haiku (claude-3-5-haiku-latest) — Fast
  • Cost per request: ~$0.0025
  • Cost to inventory 100 items: ~25 cents ($0.25)
• Claude 3.5 Sonnet (claude-3-5-sonnet-latest) — Stable Sonnet
  • Cost per request: ~$0.009
  • Cost to inventory 100 items: ~90 cents ($0.90)

Summary Recommendation:

• For everyday room items: Gemini Flash or GPT-4o Mini are the best choices. You can inventory a 1,000-item house for less than 10 cents.
• For highly specific or rare items (like art, jewelry, or specialized collectibles): Switching to Gemini Pro or Claude 3.5 Sonnet for just those groups is highly effective for maximum accuracy.