@tamastorok.eth maybe AI should evaluate answers? 😅

Builder: drawcast.xyz, mathcast.xyz, okcomputers.club, onchainbits.xyz, BA5ED Merch, basedyear.com and more | Host @ Based Budapest

Full-stack web3 developer, learn more about my projects here: https://github.com/tempe-techie

My bad, the prompt is duplicated. AI evaluation is an amazing idea!

Yeah, you could use some cheap models for that, no need for state-of-the-art models.

For example, gpt-4o-mini costs $0.15 per 1M tokens, and $0.075 for cached input of 1M tokens (I imagine a lot of your inputs will be repeated, so this lower cost cached pricing will kick in).

Or you could use Google's Gemini models, e.g. Gemini 2.0 Flash ($0.10/1M, $0.025 cached), or Gemini 1.5 Flash-8B ($0.0375/1M, $0.01 cached)

Of course you could do caching yourself, too

Or you could use Google's Gemini models, e.g. Gemini 2.0 Flash ($0.10/1M, $0.025 cached), or Gemini 1.5 Flash-8B ($0.0375/1M, $0.01 cached)

https://ai.google.dev/gemini-api/docs/pricing

Yeah, you could use some cheap models for that, no need for state-of-the-art models.

For example, gpt-4o-mini costs $0.15 per 1M tokens, and $0.075 for cached input of 1M tokens (I imagine a lot of your inputs will be repeated, so this lower cost cached pricing will kick in).

https://platform.openai.com/usage